Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios

Technische Universität Kaiserslautern Fachbereich Elektrotechnik und Informationstechnik Lehrstuhl für Integrierte Sensorsysteme Prof. Dr.-Ing. Andreas König Master Thesis Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Zoé Espejo Miralles Supervisor: Prof. Dr.-Ing. Andreas König Begin: 1. May 2012 End: 28. March 2013 Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 3 Erklärung Hiermit erkläre ich, dass ich die vorliegende Diplomarbeit selbstständig verfasst habe. Bei der Erstellung wurden keine weiteren als die angegebenen Quellen und Hilfsmittel sowie Ratschläge des Betreuers verwendet. Declaration I hereby declare that the following Master Thesis is a genuine paper. In the elaboration of this thesis, no other source apart from those cited in the bibliography was used, with the exception of recommendations and counselling received from the supervisor of the project. This text is a translation to the best of my abilities of the German legal declaration, which is binding and I consent to sign. Declaració Per la present declaro que el subsegüent Projecte Final de Carrera és un document genuí. En la elaboració d’aquest projecte no s’han fet servir altres fonts a part de les citades a la bibliografia, amb l’excepció de les recomanacions i consells rebuts per part del supervisor del projecte. Aquest text és una traducció feta amb les meves millors habilitats de la declaració legal alemanya, la qual és vinculant i consento a signar. Declaración Por la presente declaro que el subsiguiente Proyecto Final de Carrera es un documento genuino. En la elaboración de este proyecto no se han usado otras fuentes que las citadas en la bibliografía, con la excepción de las recomendaciones y consejos recibidos por parte del supervisor del proyecto. Este texto es una traducción hecha con mis mejores habilidades de la declaración legal alemana, la cual es vinculante y consiento a firmar. Kaiserslautern, den 28. März 2013 ……………………………………………………… Zoé Espejo Miralles Page 4 Report Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 5 Summary The Institute of Integrated Sensor Systems (ISE) at Technische Universität Kaiserslautern researches the design and application of intelligent, environment-aware systems using integrated, adaptive electronics and sensors. Prof. Dr.-Ing. Andreas König, chair or the ISE, started a research path focused on the automation and optimisation of cooking and food management, related to Advanced Metering Infrastructure (AmI), Ambient Assisted Living (AAL) and general home automation. These research topics are part of the Smart Kitchen or Culinary Assistance Systems scenarios developed at ISE. Although nowadays almost all disciplines make use of technology, whether it is intensively or slightly, to enhance the performance or improve the results obtained during the course of an activity, cooking seems to be a practice anchored in the past. The main objective of Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios is to add an aid tool for cooking purposes using available resources, as well as serving as a record tool, introducing technology to the kitchen environment. The Lab-on- Spoon project seeks to be innovative and focus on being a low power consumption device. The scope of the overall project is too wide for single study. It is, consequently, divided in smaller portions of work, each one focusing on certain topics, to be handled and studied by several investigators. This thesis corresponds to the first stage of the development of the project, consisting in an exhaustive study of the fitting sensors, a focus on electrochemical impedance spectroscopy and how to perform it while respecting the power consumption limitations, and a communication protocol to achieve data transfer between the microprocessor and the chip performing the impedance spectroscopy sweep. Page 6 Report Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 7 Index ERKLÄRUNG _________________________________________________ 3 SUMMARY ___________________________________________________ 5 INDEX _______________________________________________________ 7 1. GLOSSARY ______________________________________________ 9 2. INTRODUCTION _________________________________________ 11 2.1. Objectives of the project ................................................................................ 11 2.2. Scope of the project ....................................................................................... 12 2.3. Related projects ............................................................................................. 13 3. SENSORS ______________________________________________ 15 3.1. Temperature sensor ...................................................................................... 15 3.1.1. Sensor type choice ................................................................................................ 16 3.1.2. Class A Platinum RTD choice ............................................................................... 19 3.2. Acidity sensor ................................................................................................. 22 3.3. Salinity sensor ................................................................................................ 25 3.3.1. Electrochemical Impedance Spectroscopy .......................................................... 30 3.3.2. Elementary analysis of impedance spectra.......................................................... 34 3.3.3. Measuring technique and data analysis ............................................................... 40 3.3.4. EIS circuit for salinity sensor .................................................................................. 45 3.3.5. Electrode choice ..................................................................................................... 49 4. ARCHITECTURE _________________________________________ 51 4.1. Electronics ..................................................................................................... 51 4.1.1. EFM32-G890-F128 Gecko Development Kit ....................................................... 51 4.1.2. User interface tools ................................................................................................ 54 4.1.3. AD5933 ................................................................................................................... 54 4.1.4. Block diagram of the system ................................................................................. 59 4.2. Software and hardware architecture .............................................................. 60 4.2.1. Structure of the Lab-on-Spoon .............................................................................. 60 4.2.2. Structure of the EIS Sweep Sequence ................................................................. 62 4.2.3. Observations regarding code ................................................................................ 66 Page 8 Report 5. 3D PRINTING ____________________________________________ 67 5.1. CAD model .................................................................................................... 67 5.1.1. CAD models to test the 3D printer......................................................................... 68 5.1.2. CAD model for the concavity of Lab-on-Spoon.................................................... 70 5.2. Preparation of CAD models for printing ......................................................... 73 5.2.1. 3D printer tests........................................................................................................ 79 6. PROTOTYPE ____________________________________________ 80 7. EXPERIMENTS IN APPLICATION ___________________________ 84 7.1. Calibration of the system ............................................................................... 84 7.1.1. System phase calculation ...................................................................................... 84 7.1.2. Gain factor calculation ............................................................................................ 85 7.1.3. Experimental determination of calibration parameters......................................... 86 7.2. Experiments in application with electronic networks ...................................... 88 7.3. Observations during experiments in application ............................................ 92 7.4. Problems during experiments in application .................................................. 96 7.5. State after experiments in application ............................................................ 97 8. CONCLUSIONS __________________________________________ 99 9. APPRECIATIONS ________________________________________ 100 10. BIBLIOGRAPHY_________________________________________ 101 10.1. Bibliographic references .............................................................................. 101 10.2. Complementary bibliography ....................................................................... 104 11. APPENDICES ___________________________________________ 107 11.1. Appendix A: Lab-on-Spoon_main.c ............................................................. 107 11.2. Appendix B: i2c_AD5933.h .......................................................................... 109 11.3. Appendix C: i2c_AD5933.c .......................................................................... 111 Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 9 1. Glossary Lab-on-Spoon The term refers to a shortening of the title of the project (Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios), describing a device in the shape of a spoon including electronics belonging to sensor technology that turn it into a portable laboratory. Electrochemical Impedance Spectroscopy Method of characterizing electrical properties of materials and their interfaces with electronically conducting electrodes. It is measured via the current through an electrochemical cell, to which a small excitation AC potential is being applied. The mathematical analysis is done using Fourier series. I2C Inter-Integrated Circuit or “two-wire interface”. It is a world standard, widely implemented by IC manufacturers and used in various control architectures. The I2C-bus allows easy communication between components located in the same circuit board or connected by cable. Simple master/slave relationships exist between all components, and each device connected to the bus (only two bus lines are required) is software-addressable by a unique address. Nyquist plot Polar trace of the frequency answer of a system. Impedance profile Representation of the variation of impedance (Ω) with frequency. Page 16 Report Spoon in these circumstances. Stove-cooked and oven-cooked recipes, although having the potential of causing accidental damage, will be considered assuming a correct and responsible usage of the device. Combining safe cooking temperatures [4] and temperature values attainable during cooking processes [5] [6], the temperature range is determined. The minimum value of the range corresponds to that achieved by freezing processes, while the maximum value corresponds to the highest temperature achieved in oven cooking. Consequently, the sensor should be able to work, at least, in a range of [-18ºC, 260ºC]. Nonetheless, in order to allow a safety margin, the operating range of the sensor chosen should be wider. 3.1.1. Sensor type choice Several types of sensors exist in order to quantify temperature, namely thermocouples, thermistors and resistance temperature detectors. Thermocouples consist of two conductors of different materials, usually metallic alloys, and base their operational principle in the Seebeck effect. This effect is the conversion of temperature differences directly into electricity: two metals joined in two places, with a temperature difference between the junctions, create a current loop and a magnetic field due to the different responses of the metals to temperature differentials. It was discovered by physicist Thomas Johann Seebeck, and named after him [7][8]. Thermistors are a type of resistor in which the resistance varies significantly with temperature, exhibiting a large change in resistance proportional to a small change in temperature. As a temperature-sensing element, it is composed of sintered semiconductor material. Linear behaviour is assumed in most temperature ranges, and it can be increased using simple circuits [9]. Resistance Temperature Detectors (RTD) are wire wound and thin film devices that measure temperature correlating the resistance of the RTD element with temperature or, in other words, the physical principle of the positive temperature coefficient of electrical resistance of metals. RTD elements consist of a fine coiled wire of pure material, with resistance at various temperatures has been documented. Therefore, the change in resistance with temperature variations is predictable. Commonly, this material is platinum [10]. Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 17 SENSOR TYPE MATERIAL TEMPERATURE RANGE ACCURACY RESOLUTION ERROR SOURCES (+ & -) °C ±°C ºC THERMOCOUPLES J Iron-Constantan -204~1200 0,3 0,1 Cold junction compensation Deviation at Curie point Signal error and noise K Chromel-Alumel -184~1373 0,3 0,1 Cold junction compensation Deviation at Curie point Signal error and noise T Copper-Constantan -225~400 0,2 0,1 Cold junction compensation Signal error and noise E Chromel-Constantan -217~1001 0,3 0,1 Cold junction compensation Signal error and noise R Platinum-Platinum 13%/ Rhodium 0~1770 0,7 0,1 Cold junction compensation Signal error and noise S Platinum-Platinum 10%/ Rhodium 0~1769 0,7 0,1 Cold junction compensation Signal error and noise B Platinum 30%/ Rhodium- Platinum 6%/ Rhodium 481~1820 1,1 0,1 Cold junction compensation Signal error and noise N Nicrosil-Nisil 0~1260 0,3 0,1 Cold junction compensation Signal error and noise C Tungsten 5%/ Rhenium- Tungsten 26%/ Rhenium -23~2315 0,6 0,1 Cold junction compensation Signal error and noise G Tungsten-Tungsten 26%/ Rhenium 146~2318 0,6 0,1 Cold junction compensation Signal error and noise D Tungsten 3%/ Rhenium- Tungsten 25%/ Rhenium -22~2317 0,6 0,1 Cold junction compensation Signal error and noise CGI Chromel-Gold 0,07%/ Atomic Iron -273~7 0,2 0,1 Cold junction compensation Signal error and noise FeCon Iron Constantan DIN -201~901 0,2 0,1 Cold junction compensation Signal error and noise CuCon Copper-Constantan DIN -201~601 0,5 0,1 Cold junction compensation Signal error and noise THERMISTORS YSI 400 Gold-plated -41~105 0,2 0,01 Cable wire effects Self-heating Linearization error YSI 700 Gold-plated 0~100 0,2 0,01 Cable wire effects Self-heating Linearization error RTDs Pt100 Platinum -202~855 0,1 0,1 Self-heating Pt200 Platinum -201~852 0,1 0,1 Self-heating Pt1000 Platinum -204~1200 0,1 0,1 Self-heating Table 3-1: Comparison between Thermocouples, Thermistors and RTDs Page 18 Report Considering the temperature range the sensor must be able to work in, the accuracy achievable, the resolution and the minor induced errors, RTDs are the best option for the Labon-Spoon as can be stated in Table 3-1. Furthermore, platinum resistance temperature detectors are widely available through a large amount of manufacturers and are normalized by the regulations on DIN IEC 751. The DIN IEC 751 consists on a correlative table between temperature and resistance, as well as maximum deviations allowed depending on the class of the sensor. As observable in Figure 3-1, Class A Platinum RTDs have smaller tolerance values in the pursued range of operation. Consequently, a Class A Platinum RTD that completely adjusts to the Lab-on-Spoon appliance is selected, among a group of possible candidates. Figure 3-1: Tolerance values per Class for Platinum RTDs (DIN IEC 751) Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 19 3.1.2. Class A Platinum RTD choice Availability and fast delivery are factors that must be considered in the final choice of the sensors for the Lab-on-Spoon, especially with spare pieces for the future in mind. Consequently, manufacturers that provide quick customer service and whose products can be bought through their official sales representatives as well as secondary retailers are taken into account. Through Farnell | Element 14 and their own websites, Labfacility and Innovative Sensor Technology sensors are taken into account. Omega and Umwelt Sensor Technik sensors are researched through their websites. U.S. Sensor’s information is provided through their website and DigiKey. These companies offer a wide enough range of Class A Platinum RTDs to choose from, each of them with its advantages and drawbacks that must be analysed in order to perform a suitable selection for the desired appliance. In addition, Umwelt Sensor Technik is a Research & Development partner for the ISE. Regarding possible errors, given the size of the Lab-on-Spoon, cable wire effects are negligible. Self-heating errors, consisting on measurement discrepancies due to the sensor increasing its own temperature, are likely to occur and therefore must be taken into account. Linearization errors can be solved during all stages of development, in the case they were to happen. The criteria to make a final selection of the temperature sensor include size restrictions, adjustability to the operational temperature range, low consumption, low selfheating coefficient and price. The sensor selected must reach a compromise between these factors, obtaining an equilibrated performance. Page 20 Report MODEL NUMBER NOMINAL RESISTANCE TEMPERATURE RANGE DIMENSIONS (øxL / WxLxH) SELFHEATING PRICE Ω ºC mm mW/ºC € LAB FACILITY DM-508 100 -50~550 2,0x5,0 20 6,50 DM-303 100 -50~550 2,0x2,3 20 7,22 DM-312 100 -50~550 1,2x4,0 20 8,68 DM-314 100 -50~550 1,2x1,6 20 8,68 DM-310 1000 -70~600 2,0x10,0 2 7,96 INNOVATIVE SENSOR TECHNOLOGY P0K1.232.6W.A.010 100 -200~600 2,0x2,3x1,3 4 6,55 P1K0.232.6W.A.010 1000 -200~600 2,0x2,3x1,3 4 6,71 P1K0.520.6W.A.010 1000 -200~600 2,0x5,0x1,3 7 8,43 P0K1.161.6W.A.010 100 -200~600 1,2x1,6x0,8 1 13,73 P1K0.161.6W.A.010 1000 -200~600 1,2x1,6x0,8 1 14,96 OMEGA 1PT100KN1515CLA 100 -200~600 1,5x15,0 12,5 10,18 1PT100KN2515CLA 100 -200~600 1,5x25,0 12,5 17,81 1PT100KN3045CLA 100 -200~600 4,5x30,0 4,75 17,81 1PT100KN3026CLA 100 -200~600 2,6x30,0 17 23,86 UMWELT SENSOR TECHNIK FMS 2100 100 -50~400 3,0x10,0x1,3 8 UNK FMS 2101 100 -50~400 2,0x10,0x1,3 6 UNK FMS 2103 100 -50~400 2,0x5,0x1,3 6 UNK FMS 2105 100 -50~400 2,0x2,3x1,3 3,5 UNK FMS 2131 500 -50~400 2,0x10,0x1,3 6 UNK FMS 2133 500 -50~400 2,0x5,0x1,3 6 UNK FMS 2141 1000 -50~400 2,0x10,0x1,3 6 UNK FMS 2145 1000 -50~400 2,0x2,3x1,3 6 UNK FMP 2100 100 -50~600 3,0x10,0x1,3 8 UNK FMP 2101 100 -50~600 2,0x10,0x1,3 6 UNK FMP 2103 100 -50~600 2,0x5,0x1,3 6 UNK FMP 2105 100 -50~600 2,0x2,3x1,3 3,5 UNK FMP 2107 100 -50~600 1,5x5,0x1,3 8 UNK FMP 2108 100 -50~600 1,25x1,6x1,0 4 UNK FMP 2131 500 -50~600 2,0x10,0x1,3 6 UNK FMP 2133 500 -50~600 2,0x5,0x1,3 6 UNK FMP 2141 1000 -50~600 2,0x10,0x1,3 6 UNK FMP 2145 1000 -50~600 2,0x2,3x1,3 6 UNK U.S. SENSO R PPG101A6 100 -200~600 1,0x2,0x1,0 1,8 53,02 PPG102A6 1000 -200~600 1,0x2,0x1,0 1,8 53,02 Table 3-2: Characteristics for analysed sensors Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 21 Considering the minimization of the self-heating error, only sensors with a self-heating coefficient equal or inferior to 8 are considered. With the purpose of being as accurate as possible and gain resolution in the processing of data, narrower temperature ranges closer to the cooking range, while still allowing a safety margin, are preferred. For these reasons, Umwelt Sensor Technik offers sensors that satisfy the needs of the Lab-on-Spoon. In this stage of the project, for the sake of trailblazing, the temperature sensor is left aside, as the complexity of the project in the time span given requires compromises to be achieved in order to focus in other areas of the investigation that provide more interesting results and innovative appliances of the current state of the art. The ISE has readily available a platinum temperature sensor from a previous project. This sensor has provided very satisfactory results, with a high accuracy and easy handling. It is packed in a small die, with a nominal resistance of 10000 Ω, making it suitable for the purposes and sizing issues of the studied application. It could handily be installed without disturbance of the electronics and their already achieved disposition on the project. It is, thus, advisable to continue using said sensor in future implementations of the Lab-on-Spoon, as it has previously been tested and it would mean economic savings on the overall cost of the project. Page 22 Report 3.2. Acidity sensor Acidity is a magnitude with no objective description in cooking that is largely unexplored. Despite this fact, acidity is extremely relevant in the success of a recipe due to the taste it gives to it, and a determining factor when sorting rotten or spoiled food. When there is a chemical reaction between food (considered stimuli) and the taste buds located in the human tongue (receptors), the sensation of taste is produced. Along with smell and trigeminal nerve stimulation (texture, pain and temperature), flavours, the sensory impressions of food, are determined [11]. The sour taste is one of the five basic tastes that the taste buds of the tongue are able to detect. The sensors on the taste buds detect, through ionic channels, the hydronium (H3O+) formed when acids are present in water. The perception of a sour taste leads to reactions of aversion towards the source of the taste, due to many dangerous or non-edible foods having this characteristic taste. This aversion is a natural defence against incidental ingestion of poisonous substances [12]. Consequently, an otherwise perfectly cooked recipe could go to waste if it is too sour, as the human brain would interpret it as a potential harm and would associate it as something bad. Furthermore, if the food is in an unhealthy condition, the acidity in it would give away said condition. With acidity correctly sensed, sourness and rottenness can be detected and a way of solving these problems can be figured out. Given the actual state of the art, the acidity level in food can be determined using pH sensors that do not require huge equipment nor are out of budget. In chemistry, pH is a measure of the activity of the (solvated) hydrogen ion, and the acronym stands for “minus decimal logarithm of hydrogen” [13]. Since the scale is logarithmic, pH is a dimensionless quantity. Pure water has a pH very close to 7 at 25°C. Solutions with a pH less than 7 are said to be acidic and solutions with a pH greater than 7 are basic or alkaline. The pH scale is traceable to a set of standard solutions whose pH is established by international agreement. These solutions are known as buffer solutions according to the IUPAC, and have known H+ activity at 25ºC, as well as a correction factor to be applied for other temperatures. Following the International Standard ISO 31-8 for precise measurement of pH, a galvanic cell is set up to measure the electromotive force (e.m.f.) between a reference electrode and an electrode sensitive to the hydrogen ion activity when they are both Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 23 immersed in the same aqueous solution. A glass electrode is used as the hydrogen-ion selective electrode. A combined glass electrode has an in-built reference electrode, such as silver chloride electrode or calomel electrode. Two or more buffer solutions are used to calibrate the measurement, accommodating the slightly variation from the ideal value of the slope. When pH levels are below 2.5 or above 10.5 approximately, special procedures are required due to the electrode potentials being affected by ionic strength variation or sensitivity to the concentration of cations such as Na+ and K+ in the solution. Current state of the art allows to measure pH through the use of ISFET [3], an ionsensitive field-effect transistor used for measuring ion concentrations in a solution, used as the gate electrode. An ions sheath provokes a difference of potential between the substrate and the oxide surfaces of the ISFET. For the Lab-on-Spoon, said solution is the food being prepared or being tested for rottenness. The site binding model describes the equilibrium between Si-OH surface sites and H+ ions, and is the responsible mechanism for the oxide surface charge. The hydroxyl groups coating an oxide surface (typical gate materials are SiO2, Si3N4, Al2O3 and Ta2O5) can donate or accept a proton and behave in an amphoteric way. In other words, these groups can react as an acid as well as a base, depending on the pH of the solution they are in. The source and drain of the ISFET are constructed in the same manner as it would be in a MOSFET, normally as a NMOS technology of five levels. The gate, as stated before, is a layer of material sensitive to the variation of pH in the solution [14]. A barrier sensitive to hydrogen ions and a gap to allow contact between the sensitive barrier and the tested substance is used to separate the gate electrode from the channel. The threshold voltage depends on the pH of the substance. The measurement principle is the modulation of the channel in the ISFET through the voltage difference between the substance and the gate of the ISFET. Page 24 Report Current state of the art regarding ISFET pH sensors makes it difficult to find nude sensors, as the vast majority of the available options are already included in sophisticated readily prepared probes. Visitors in the ISE at Technische Universität Kaiserslautern have provided samples of nude sensors. Unfortunately, due to the complexity of its implementation and the lack of information regarding the sample, it is finally decided to set aside the sensing of pH. Investigation about pH sensors is, therefore, left for future continuers of the Lab-on- Spoon project. Figure 3-2: Schematic of an ISFET, as described by the IUPAC Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 25 3.3. Salinity sensor Saltiness is an enhancer of the overall taste in any recipe. An enhancer is a substance that, used in the concentrations normally adequate for food processing, does not provide its own flavour to the recipe although it does have one, but boosts that of the other found components in the recipe. Enhancers also have an influence on the full-bodied sensation when savouring the dish and also in the viscosity, increasing both. Soups and sauces benefit specially from these properties of saltiness, but salt is used in a wide range of recipes. As well as acidity, salinity is one of the five basic tastes the human tongue is able to identify. This taste is produced by the presence of sodium ions and other ions of the alkali group, although when these ions get far from sodium the salty sensation decreases. The detection of saltiness is done through ionic channels capable of detecting said soluble ions. The salty sensation is more perceivable when the ions have a low molecular weight. Lithium and potassium ions resemble closely in size sodium ions, giving a similar saltiness. On the contrary, rubidium and caesium are larger than sodium ions, giving an overall sensation of bitterness [12][15]. Table or marine salt (NaCl) is one of the most used seasoning in the kitchen. It becomes an indispensable element during cooking due to its properties as flavour enhancer. Moreover, when eaten, salt gives the need to ingest more food. It can also be used as a preservative, in order to prevent fish or meat from rotting. Due to its common usage, the saltiness of a substance is rated relative to sodium chloride (NaCl), which has an index of 1. Salt substitutes such as potassium chloride (KCl) have a saltiness index of 0.6 [11]. Table salt has been proven to be of vital importance for the human body, due to its influence in the digesting process and its job as a maintainer of pressure in corporal fluids, blood pressure and acid equilibrium in the human body. If a person does not ingest a minimum quantity of sodium, symptoms like fatigue and confusion appear, with a possible dangerous lead to more serious seizures or even coma, and the most severe case of death. Although it is necessary for the organism, an excessive consumption can be detrimental to health. It can cause fluid retention, producing an increase of the muscular volume, and it is damaging to the health of people with a tendency of having high blood pressure. Such damages could be heart disease, stroke or kidney failure. Some studies also reveal a relation between excessive salt consumption with stomach cancer, Ménière’s disease or pica disorder [16]. Page 32 Report As previously stated, impedance can be described as the quotient between voltage and current (   j eZZjZ I V Z·)()()(      ). Figure 3-6: Flow diagram for the measurement and characterization of a materialelectrode system Figure 3-7: Representation of Z(ω) MATERIAL-ELECTRODE SYSTEM IS EXPERIMENT Ze(ω) THEORY PLAUSIBLE PHYSICAL MODEL MATHEMATICAL MODEL Zt(ω) EQUIVALENT CIRCUIT Zec(ω) CURVE FITTING (CNLS) SYSTEM CHARACTERIZATION Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 33 Nyquist plots, obtained when – Z’’(ω) is plotted against Z’(ω), make all the necessary circuit parameters available from them, as the fundamental laws which connect charge and potential and which define the properties of linear systems are unchanged in passing from electronic to ionic materials. The impedance spectrum is constructed putting the real part as well as the imaginary part of Z(ω) in a same graphic, making them a function of frequency ƒ [23][24]. It is from the resulting structure of the Z(ω) vs. ω or ƒ response that information about the electrical properties of the full electrode-material system is derived. Both solid and liquid electrochemical systems tend to show strong nonlinear behavior, especially in their interfacial response, when applied voltages or currents are large. As long as the applied potential difference amplitude Vm is less than the thermal voltage VT (about 25 mV at 25ºC), it can be shown that the basic differential equations which govern the response of the system become linear to an excellent approximation. Therefore, if the applied amplitude Vm is appreciably smaller than VT, the system will respond linearly. gecharrotonPe antconstsBoltzmannk antconstsFaradayF etemperaturAbsoluteT antconstGasR e Tk F TR VT       ' ' ·· (Eq. 3.10) Figure 3-8: Pseudo-linearity when Vm <<VT Page 34 Report 3.3.2. Elementary analysis of impedance spectra An electrochemical reaction that takes place on the interface between an electrode and an electrolyte is described using an electrical equivalent circuit (EEC) as a model. Via EIS, the use of said model for the reaction at the electrified interface is examined, due to the direct connection than often exists between the behaviour of a real system and that of an idealized model circuit consisting of discrete electrical components. Regardless of the thoroughness of a measurement, the current flowing at an electrified interface due to an electrochemical reaction always contains non-faradic components. This reaction follows the typical chemical reduction equation. antteducRR dtransferreelectronsofNumberne OxidantO RneO       (Eq. 3.11) Figure 3-9: Electrified interface and the reactions that take part in it (IHP/OHP ≡ Inner/Outer Hemholtz Plane) Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 35 The electron is transferred across the electrified interface. The charge transfer leads to both faradic and non-faradic components. The faradic component arises from the electron transfer via a chemical reduction reaction across the interface by overcoming an appropriate activation barrier, namely the polarization resistance (Rp) along with the uncompensated solution resistance (Rs).The non-faradic current results from charging the double-layer capacitor (Cd ). When the charge transfer takes place at the interface, the mass transports of the reactant and product determine the rate of electron transfer, which depends on the consumption of the oxidants and the production of the reductant near the electrode surface. The mass transport of the reactants and the products provides another class of impedance (ZW). The activation barrier at any potential is represented by the polarization resistance (Rp), but the barrier becomes the charge-transfer resistance (Rct), at the standard electrode potential. The electrified interface can be represented by an EEC [24]. It is called the Randles equivalent circuit, and it describes the response of a single-step charge-transfer process with diffusion of reactant and/or products to the interface. Resistances represent conductive paths (bulk conductivity of the material, chemical step associated with an electrode). Capacitances and inductances are associated with adsorption and electrocrystallization processes at an electrode. They are all considered as lumped-constant quantities, involving ideal properties. Some cell elements and cell characteristics that are potential contributors to the system’s EIS spectrum include Electrode Double Layer Capacitance, Electrode Kinetics, Diffusion Layer and Solution Resistance. Figure 3-10: EEC model of the electrified interface Page 36 Report In the study of a system, two different levels of description can be achieved. The most fundamental level of description consists on atomistic or microscopic models, trying to provide an accurate description of the motions of individual charge-carrying particles in the system. Less detailed, but widely used, are equivalent circuit models. In them, there are hypothetical electrical circuits, consisting of elements with well-defined electrical properties. Between these two levels of description, multiple analysis techniques can be found, ranking from more fundamental levels to less detailed levels. The most known is the continuum level. In order to analyze EIS spectra, the most common method is equivalent circuit modelling. A cell is simulated incorporating the elements from the Randles model, and they are used to describe the response of the system to a range of possible signals. Commonly, an educated guess is done as a first step to assigning an EEC to the system, predicting the system elements taking part in the cell’s impedance. Each element in the model has known impedance behaviour, and said impedance depends on the element type and the value of the parameters that characterize the element. Numerous theoretical models have been developed to explain and predict the behaviour of electrochemical systems and to guide the design of systems with desired characteristics. As stated before, one of the biggest inconvenient of the EIS analysis is the ambiguity the equivalent circuits may present. Only the simplest EECs can be said to be unique. The three circuits in Figure 3-11 have different elements, but they all present the same impedance at every frequency. They serve as an example of this inconsistency that must be taken into account when establishing a suitable EEC for the studied system. Figure 3-11: Example circuits with same impedance Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 37 A criterion to select the most suitable EEC in case of ambiguity must be agreed on, especially in cases when element estimates for the different circuits are quite different. As stated previously, continuity and knowledge of the physical processes involved in the system must be taken into account. This should give a good estimate of the final EEC. A useful criterion is comparison with predictions of a physical model: there might be an option that produces simpler expressions for the elements of one of the circuits than for the others. In the case of EIS studies, the expected physical model is the Randles model. Finally, simplicity criterion must be applied. In the case of equality between good fits, the circuit with the smallest number of elements should be used. When electrode separation, temperature, oxygen partial pressure or other physical conditions are changed, it is expected to encounter a change in some or all the fitting parameters. Therefore, the selected circuit would generally be preferred to be the one in which the changes are least, simplest and/or closest to theoretical expectations. Following this criteria, one can often reach an almost unambiguous choice of the most suitable fitting circuit, out of the considered to use. Electrolytic cells or samples are always finite in extent. The electrical response exhibits two generic types of distributed response, requiring the appearance of distributed elements in the equivalent circuit. The first type of distributed response, diffusion, appears due to the finite extent of the system, regardless of homogeneity and space-invariance of the properties. It can lead to a distributed circuit element, the impedance of which might not be exactly expressed as the combination of a finite number of ideal circuit elements. As the electrodes are of macroscopic dimensions, the total macroscopic current flowing in response to an applied static potential difference is the sum of a very large number of microscopic filaments originating and ending at the electrodes. The individual contributions to the total current are all different if there is roughness in the electrodes and/or the bulk properties of the material are not homogeneous. This leads to a distributed resistance or conductance, that is, many differential elemental resistances or conductance. When small-signal frequency and time dependence is considered, the result may be described in terms of a distribution of relaxation times, leading to frequency-dependent effect. These may, at least in an approximate way, be described through the use of certain simple distributed circuit examples. The first distributed element introduced into electrochemistry was the infinite length Warburg impedance, in 1899. Nevertheless, physical experiments are done in a finite-length region (equivalent to a finite-length, shorted transmission line). The solution for the diffusion of particles in such a region was first presented by Llopis and Colon in 1958 for the supported situation, where the finite length considered was the thickness of the Nernst diffusion layer, appropriate for a stirred electrolyte or a rotating electrode. It results, therefore, in the finite-length Warburg impedance. It appears in supported situations (sometimes also in unsupported ones) and exhibit a characteristic θ=45º lines in the Z* plane. Page 38 Report                        RROO W DCDCAFn TR D j jZ · 1 · 1 · 2··· · ·)·tanh1·(· **22 2 1 2 1    (Eq. 3.12) ZW is the finite-length Warburg impedance (sometimes also noted as Z0) and σ is the Warburg coefficient. The rest of the parameters: ntreductatheoftcoefficiendiffusionD oxidanttheoftcoefficiendiffusionD bulktheinntreductaofionconcentratC bulktheinoxidantofionconcentratC electrodetheofareasurfaceA ntconstasFaradayF involvedelectronsofnumbern etemperaturT ntconstagasR speciesngdiffusitheoftscoefficiendiffusiontheofvalueaverageD thicknesslayerdiffusionNernst R O R O            * * '  Often, though, approximate straight-line behaviours over a limited frequency range with θ≠45º can be found. This effect is called Constant Phase Element or CPE, due to the fact that its phase is independent of frequency. The CPE exhibits no transition from intensive to extensive behaviour as the frequency decreases. Then the frequency response of Z’ and Z’’ is no longer proportional to ω-½, but to some other power of ω.   )·( 1 0jA ZCPE  (Eq. 3.13) A0 and ψ are parameters temperature-dependant, and 0≤ψ≤1. For ψ=1, it describes an ideal capacitor. For ψ=0, it describes an ideal resistor. Constant phase response is generally thought to arise with 0<ψ<1 from the presence of imperfections, such as non-homogeneities, in the electrode-material system, or from the nonuniform diffusion whose electrical analogue is a non-homogeneously distributed RC transmission line. It appears in the majority of experimental data, both on solid and liquid electrolytes, but can only be well approximated over a finite range of frequency. When said Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 39 frequency is sufficiently low or high, it turns to be physically unrealizable. Therefore, at frequency extremes, the response must be deviated from the CPE type of response in order to produce a realistic, physically realizable response. In order to mathematically describe the Randles EEC, apart from defining ZW and ZCPE, a generalist vision of resistors, inductors and capacitors is necessary. The following table shows common electrical circuit elements and its characteristic equations, as well as their impedance once the Fourier transformation is applied. SYMBOL EQUATION Z(ω) RESISTOR RIU · RZ  INDUCTOR dt di LU · LjZ ··   CAPACITOR dt dU CI · C··j Z  1  The frequency-dependant impedance at the interface, noted as Z(ω), can be described with a mathematical expression related to the EEC model in Figure 3-10. Angular frequency ω is related to frequency ƒ via the equality ω = 2πƒ. Table 3-3: Electrical circuit elements )()( )()1( )1()( )()1( )( 2 2 1 2 22 2 1 2 1 2 1 2 2 1 2 2 1 2 22 2 1 2 1       ZjZ RCC CRC j RCC R RZ pdd dpd pdd p s                        (Eq. 3.14) Page 40 Report This expression, nonetheless, might be too complicated to operate with. Therefore, simplifications for low frequencies (ω→0) and high frequencies (ω→∞) can be defined. )()()2()( 0 2 2 1 2 1   ZjZCjRRZ dps       (Eq. 3.15) )()( 11 )( 22 22 2 22 22       ZjZ RC RC j RC R RZ pd pd pd p s                 (Eq. 3.16) 3.3.3. Measuring technique and data analysis The first methods used in electrochemical studies related to EIS involved the processing and analysis of analogue signals in the frequency domain or in the time domain. In the first case, impedance measurements are performed using a small-amplitude sinusoidal excitation with frequency as the independent variable. In the second case, time is the independent variable, and the impedance as a function of frequency can be extracted by time-to-frequency conversion techniques (Laplace or Fourier transformation). Characteristically speaking, frequency domain methods use analogue techniques, while time domain methods use digital-processing techniques. Nowadays, with the technological advances involving digital computers, digital processing is the clear trend in the synthesis and analysis of sinusoidal signals. This is due to the purely mathematical advantage of digital data processing over its analogue counterpart, as a far wider range of mathematical computations can be performed in the digital mode. Laplace and Fourier transformation involve the recording of the perturbation and response in digital form in the time domain before signal processing in either software or hardware. Therefore, the accuracy of the transformation is critically dependant on acquired data records having the desired characteristics of length and sampling frequency. The essential operation is the conversion of the value of an analogue signal into a binary word whose magnitude is proportional to the signal being sampled, characterized by sampling and quantization. Sampling can be defined as the extraction of a discrete signal, consisting in a set of values at a point in time, from a continuous signal. In other words, the discrete signal is performed by measuring the value of the continuous function every T seconds (sampling interval). The sampling frequency or sampling rate (ƒs) can be defined as the number of Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 41 samples obtained in one second, and therefore as the inverse of the sampling interval (ƒs=1/T). The sampled signal is obtained applying an ideal low-pass filter with a sequence of Dirac delta functions inputs. These functions are modulated or multiplied by the sample values. As the time interval between adjacent samples is a constant, the sequence of delta functions is a Dirac comb. The sampled signal or modulated Dirac comb is equivalent to the product of said Dirac comb function with the continuous signal. Quantization is the process of mapping a large set of input values to a smaller set, and it is an inherently non-linear and irreversible process due to the process being a many-to-few mapping [25]. In the case of EIS processing, it is achieved through an ADC (analogue to digital converter) and its defining characteristics. In the case of the Lab-on-Spoon, the ADC is internal to the microprocessor used in the application and presents Successive Approximation Register architecture (SAR), with a resolution of up to 12 bits at up to one million samples per second. This kind of architecture allows high speed and flexibility, although it is precision expensive and might be susceptible to noise. Nevertheless, it is one of the best alternatives currently available, even with its shortages. The microprocessor used on the Lab-on-Spoon is further described in upcoming section 4.1.1. Figure 3-12: Signal sampling representation Page 48 Report It is important to respect the settling time upon programming the registers, specially the control register. If not, the AD5933 is not set up properly, creating conflicts in the reading of data and the DFT conversions. It is preferable to delay the start of the sweep and ensure that the chip has been initialized with the start frequency without issues, as the addition of such delay would not affect the overall fast performance (the user would see it as part of the wakeup sequence, understanding that there is a waiting time between start up and usage). Figure 3-18: Frequency Sweep Flow Chart for AD5933 Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 49 3.3.5. Electrode choice One of the limitations on the exact correspondence between equivalent circuits and electrochemical systems is the effect of geometry on the current distribution. Therefore, the geometry of the electrode must be constant and known during the measuring process. Electrodes are mounted in conductivity cells, leaving a known distance between them, in a shape that can be fitted in the desired application. In the case of the Lab-on-Spoon, it is fitted in the part in contact with the analysed substance. When electrodes are too big, the measurement process could turn to a destructive process. In other words, the sample needs to be discarded after the electrode is used to measure the impedance. Furthermore, measurements cannot be conducted on the inside of the sample. In consequence, in order to avoid these problems, the electrode must be of small dimensions. This also makes feasible piercing through the sample and acquiring internal measurements without having to discard it afterwards. With small electrodes, the measurement system is non-destructive. In order to do accurate measurements that do not require subsequent disposal of the sample and/or contamination of the same, an ion-selective electrode is the optimal kind for the Lab-on-Spoon application. They are defined as electro-analytical sensors with a membrane whose potential indicates the activity of the ion to be determined (determinand) in a solution (analyte). In the case of study, sodium ions and other ions from the alkali group are to be the determinand. Said membranes consist of liquid electrolyte solutions or solid or glassy electrolytes, the electron conductivity of which is usually negligible under the conditions of measurement. This is especially useful in applications where it is necessary to know that a particular ion is below a certain concentration level. It has been also proved that ion-selective electrodes do not affect the test solution, a feature that the Lab-on-Spoon application requires. Furthermore, they are portable, suitable for direct determinations and are relatively inexpensive. Recent plastic-bodied all-solid-state or gel-filled models are very robust and durable under both field and laboratory circumstances, and they are not affected by the colour of the sample nor its turbidity. Under favourable conditions and considering interfering ions are not a problem, ion-selective electrodes can be used in aqueous solutions fairly quickly and easily. Since the EIS requires fast setting times, this kind of electrodes result in the most suitable. Crystal membranes can operate in the range 0ºC to 80ºC, while plastic membranes can do so between 0ºC to 50ºC [28]. Due to the Lab-on-Spoon’s temperature requirements, high temperature insertion type electrodes must be taken into account. The ion-selective electrode is housed in a durable thermoplastic (PAS) body for use with higher temperatures. The use of double porous PTFE liquid junctions with matched viscosity electrolytes provides a reference cell that allows the Page 50 Report usage of the electrode in heated environments [29]. However, these increases the final price of the electrode, getting out of the planned budged for the Lab-on-Spoon project. A compromise can be met by the usage of a regular two-plate cell placed in the concavity of the spoon in a position that allows knowing, as accurately as possible, the surface of each plate and the difference between them. Although more precise results would be welcome and attainable with the use of ion-selective electrodes, the application of the Labon-Spoon can operate with less exact values, due to the calibration with standard saline solutions and posterior usage of the obtained curve to determine the salinity of the sample. Considering the electrodes are in constant contact with the sample, and that said sample is not afterwards disposed, the chosen material needs to be non-toxic and nonintrusive with food. The best candidates for this purpose would be platinum plates and stainless steel plates. The first are discarded due to their price and difficulty to be soldered using regular laboratory soldering equipment. The second are determined to be the cell in use, due to the simplicity to obtain them at reasonable prices and the possibility to perform soldering on them in a laboratory and/or academic environment. In order for the cell to work properly, the copper wires that attach to the AD5933 are stripped at the tips and soldered to the plates with silver solder. It is also required to cover the rear face of the plates with epoxy resin, for isolation and protection [18]. Figure 3-19: Electrode diagram (single plate) Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 51 4. Architecture 4.1. Electronics 4.1.1. EFM32-G890-F128 Gecko Development Kit As it has been stated before, the Lab-on-Spoon project seeks to be as low-power as possible. In the same way one of the criteria to choose the sensors was their consumption, the micro-controller must follow this directive. The ISE at Technische Universität Kaiserslautern acquires pieces of hardware for at the moment on-going projects, or that might be interesting towards the development of future ones. In an attempt to optimize the resources available, a suitable device is selected from the aforementioned pieces of hardware, respecting the consumption and speed premises. Prof. Dr.-Ing. Andreas König found suitable to use the EFM32-G890-F128 Gecko Development Kit produced by Energy Micro. As its own advertisement says, this sort of development kits based on the EFM32 has “the world’s most energy friendly microcontroller”. It is a micro-controller well suited for battery operated applications and other systems that require a high-performance and a low-energy consumption. This is possible due to the use of the powerful 32-bit ARM Cortex-M3, innovative low energy techniques, short wake-up time from energy saving modes, and a wide selection of peripherals. This family of microcontrollers (EFM32) outperforms other available 8-, 16-, and 32-bit solutions. Figure 4-1: Block Diagram of EFM32 Page 52 Report The user can connect external devices to the micro-controller using the pin headers in the prototype board (located on the right of the development kit). The MCU board (located on the top left of the development kit) contains the EFM32, as well as an easily accessible LCD screen. In the demonstration modes, the LCD screen displays the different energy modes. A TFT screen (located on the bottom left of the development kit) permits easy user access to the EFM32: displaying of energy monitoring, flashing of the unit... It is possible to reset the device through two different buttons. One of them resets the development kit board in its entirety, the other resets only the MCU board and, thus, its components. The power switch is easily turned on and off, but is located far enough from the most manoeuvred parts of the development kit board, preventing accidental switching. The EFM32-G890-F128 Gecko Development Kit includes a large amount of features, including the aforementioned ARM Cortex-M3 CPU platform (running up to 32 MHz), a flexible energy management system, 128 kB of Flash, 16 kB of RAM memory, up to 90 general purpose I/O pins, an 8 channel DMA controller, an 8 channel peripheral reflex system for autonomous inter-peripheral signalling, an external bus interface for up to 64 MB of external memory mapped space, an integrated LCD controller for up to 4x40 segments, several communication interfaces, timers, counters, ultra low power precision analogue peripherals, an ultra efficient power-on reset and brown-out detector, and a 2-pin serial wire debug interface. Its working temperature range is between -40ºC and 85ºC, and the power supply can be set between 1,8V and 3,8V. Its autonomous peripherals are energy efficient and fast. It includes high overall chip- and analogue integration. Figure 4-2: EFM32-G890-F128 Gecko Development Kit Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 53 The DBG – Debug Interface is used to program and debug EFM32G devices. It makes it easy to reprogram and update the system in field, and allows debugging with minimal I/O pin use. The Cortex-M3 supports advanced debugging features. EFM32G devices include hardware debug and programming support through a 2-pin serial-wire debug interface. In addition there is also a 1-wire Serial Wire Viewer pin which can be used to output profiling information, data trace and software-generated messages. The systems internal and external state can be examined with debug extensions supporting instruction or data access break- and watch points. The I2C – Inter-Integrated Circuit Interface is a module that provides an interface between the MCU and a serial I2C bus with the lowest energy consumption possible, capable of acting as both master and slave, and supporting multi-master buses. Transmission rates can vary from 10 kilo-bit per second to 1 Mega-bit per second, for it supports standard-mode and fast-mode speeds. It allows a precise control of the transmission process, as well as close to automatic transfers. The UART – Universal Asynchronous Serial Receiver and Transmitter is a very flexible serial I/O module that allows efficient communication with a wide range of external devices frequently used in embedded systems. The GPIO – General Purpose Input / Output is used for pin configuration and direct pin manipulation and sensing, as well as routing for peripheral pin connections. The GPIO pins are organized into ports with up to 16 pins each, which can be individually configured as output or input, as well as more advanced configurations (open-drain, filtering or drive strength). Easy to use and highly configurable pins fit many communication protocols and maximize software control overhead, with the flexible routing helping to ease PCB layout. Although not used in this stage of the Lab-on-Spoon project, the EFM32-G890-F128 features an ADC – Analogue to Digital Converter, consisting on Successive Approximation Register (SAR) architecture, with a resolution of up to 12 bits at up to one million samples per second and 8 external input channels; as well as a DAC – Digital to Analogue Converter, fully differential rail-to-rail, with 12-bit resolution or accuracy at up to 500 kilo-samples per second, designed for low energy consumption while providing an excellent performance, with two single ended output buffers that can be combined into one differential output. These features need to be taken into account, as they are necessary for the implementation of future stages of the Lab-on-Spoon project, and thus included in the block diagram of the system. Page 54 Report 4.1.2. User interface tools Energy Micro provides an innovative solution to help the user be updated regarding the Gecko Development Kit. This tool, called Simplicity Studio, provides the user with the latest documents, examples, firmware and software necessary to warrant the best performance. It is automatically kept up-to-date. With a single click, it is possible to access interesting gear, such as commander and designer tools, an energy profiler and source code libraries. There is also a product selector that helps increase the speed of the MCU selection process. Included with the Gecko Development Kit box is an installation CD for development tools from IAR Systems. Specifically, it includes IAR Embedded Workbench for ARM, an integrated development environment and optimizing C/C++ compiler for ARM microcontrollers. Its hardware debugging support includes J-Link and J-Trace, which support the Cortex-M3 processor in the EFM32-G890-F128 Gecko Development Kit. It provides device support on several levels: core support (instruction set support in compiler, assembler, linker and debuggers), header/DDF files (peripheral register names in C/assembler source and debugger, as well as device setup configuration files), flash loader (for on-chip flash or offchip EVB flash) and project examples (varying from simple to fairly complex applications). Directly mounted on the EFM32-G890-F128 Gecko Development Kit is the J-Link device by SEGGER, for development and production purposes. It is a USB powered JTAG emulator that supports a large number of target CPU cores, including Cortex-M3, with which it can communicate at high speed. It is based on a 32-bit RISC CPU. 4.1.3. AD5933 The AD5933 chip, with its frequency generator, allows an external complex impedance to be excited with a known frequency. The response signal is sampled by its internal ADC. A DSP engine processes a discrete Fourier transform (DFT) of the sample, returning a real and imaginary data-word at each output frequency once the algorithm is over. It features an internal system clock option, which will be used in the application of the Lab-on-Spoon, of 16,776 MHz and its power supply operation ranges between 2,7 and 5,5 V. The frequency sweep capability is programmable with serial I2C interface. It is very suitable for coordinate use with the EFM32-G890-F128 Gecko Development Kit, as this last can provide a direct voltage supply of 3,3 V and communicate data using the I2C interface. Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 55 The AD5933 also includes a temperature sensor on-chip. It is a 13-bit digital temperature sensor, where the 14th bit acts as the sign bit, to accurately measure the ambient device temperature. The measuring temperature range is between -40ºC and 125ºC. However, at 150ºC the structural integrity of the device starts to deteriorate when operated at voltage and temperature maximum specifications. Therefore, the chip needs to be kept at a reasonable distance from the measured sample if it is hot, by means of physical distance or good thermal isolation. The measurement of the temperature sensor can be used to force power-down mode on the AD5933 in future implementations of the Lab-on-Spoon, to prevent it from malfunctioning and breakage if the operating temperature is too high. The AD5933 has a finite frequency response, resulting in an error in the impedance calculation over a frequency range due to the gain factor varying. To minimize this error, the frequency sweep should be limited to as small a frequency range as possible. The output frequency range of the chip oscillates between 1 kHz and 100 kHz, with an output frequency resolution of 27 bit (<0,1 Hz, achievable through the use of Direct Digital Synthesis or DDS techniques). DDS is a method of producing an analogue waveform by generation a time-varying signal in digital form and then performing a digital-to-analogue conversion. It can offer fast switching between output frequencies, fine frequency resolution and operation over a broad spectrum of frequencies. Systems that produce waveforms by means of DDS have been proven to have low cost and power consumption levels. As stated before, a low power consumption is an important requisite for the Lab-on-Spoon application. The impedance measurement range is from 1 kΩ to 10 MΩ, but it is capable of measuring from 100 Ω to 1 kΩ by using additional circuitry. With it, there is an increase in signal current flowing through the impedance, resulting in the sensing process being able to carry on normally. Page 56 Report The packaging of this chip consists on a 16-lead SSOP (shrink small-outline package), as described in the following table, reproduced from the data-sheet. Pin No. Mnemonic Description 1, 2, 3, 7 NC No connect. 4 RFB External Feedback Resistor. Connected from Pin 4 to Pin 5 and used to set the gain of the current-to-voltage amplifier on the receive side. 5 VIN Input to Receive Transimpedance Amplifier. Presents a virtual earth voltage of VDD/2. 6 VOUT Excitation Voltage Signal Output. 8 MCLK The master clock for the system is supplied by the user. 9 DVDD Digital Supply Voltage. 10 AVDD1 Analogue Supply Voltage 1. 11 AVDD2 Analogue Supply Voltage 2. 12 DGND Digital Ground. 13 AGND1 Analogue Ground 1. 14 AGND2 Analogue Ground 2. 15 SDA I2C Data Input. 16 SCL I2C Clock Input. The communication between the microprocessor and the AD5933 chip is accomplished by reading from and/or writing on the registers of the chip via I2C interface. Register Name 0x80 Control 0x81 0x82 Start frequency 0x83 0x84 0x85 Frequency increment 0x86 0x87 0x88 Number of increments 0x89 0x8A Number of settling time cycles 0x8B 0x8F Status 0x92 Temperature data 0x93 0x94 Real data 0x95 0x96 Imaginary data 0x97 Table 4-1: Pin configuration and descriptions for AD5933 Table 4-2: Register map for AD5933 Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 57 The control register consists of 16 bits that set the chip control modes. The four most significative bits (that is, register address 0x80) are decoded to provide control functions, such as performing a frequency sweep, powering down the part, and controlling various other functions defined in the control register map. The start frequency is a 24-bit word programmed to the on-board RAM as a hexadecimal code calculated from the required start frequency output from the DDS and the master clock frequency. 27 2· 4                  MCLK FrequencyOutputequiredR CodeFrequencyStart (Eq. 4.1) The frequency increment is also a 24-bit word programmed to the on-board RAM as a hexadecimal code as a result of a very similar formula to that used for the start frequency. 27 2· 4                  MCLK IncrementFrequencyequiredR CodeIncrementFrequency (Eq. 4.2) The number of increments is a 9-bit word representing the number of frequency points in the sweep. The maximum number of points that can be programmed is 511. The number of settling time cycles is a register that determines the number of output excitation cycles that are allowed to pass through the unknown impedance before the ADC is triggered to perform a conversion of the response signal or, in other words, the delay between a command to allow a frequency sweep and the time an ADC conversion starts. It is represented by a 9-bit word that can be increased by a factor of 2 or 4. The maximum number of output cycles that can be programmed is 2044. The status register is used to confirm that particular measurement tests have been successfully completed: valid temperature measurement, valid real and imaginary data, completed frequency sweep. Page 64 Report Figure 4-6: Flow chart of the programming stages for the AD5933 Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 65 To perform a frequency sweep, function AD5933_Sweep is needed. It reads the status register to assure the real and imaginary data stored is a valid piece of data. If it is, the hexadecimal value is displayed on the LCD screen and also saved via UART communication. This process repeats itself for every frequency increment, until the moment the status register polls that the frequency sweep is complete. The AD5933 is placed into power-down mode and, after a certain delay; the whole system is put to sleep, as stated in Figure 4-4. Figure 4-7: Flow chart of the frequency sweep process for the AD5933 Page 66 Report 4.2.3. Observations regarding code The functionalities related to the visual aids, namely writing state messages on the LCD screen and lighting up LEDs, as well as displaying the real and imaginary data results, are only to be included in this first stage of the Lab-on-Spoon project for debugging purposes. As the project is evolved, these aids are bound to disappear, as the completed version does not feature an LCD screen or LEDs. However, the base idea of having an easily discernible display of the state the sensing sequence is in could be recycled and included as text messages to be written as part of the contents of the DataLogger (data to be sent to a computer using UART communication), making the interface more user friendly. In the intermediate stage between working with the EFM32-G890-F128 Gecko Development Kit and the dedicated PCB, it would be advisable to set these functionalities as conditional parameters in the compilation of the source files. This way, depending on the device in use, the visual aid functionalities could be included or not in the program. A complete version of the code written for the Lab-on-Spoon application, which has been briefly described, can be found in the Appendices. The structures previously described can be thoroughly followed and examined for its complete understanding. Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 67 5. 3D printing 5.1. CAD model The Lab-on-Spoon has a very distinctive geometry, as the sensors are integrated in the concavity of a spoon-like device. In order to make it ready to use, CAD models are designed to satisfy the needs of the appliance. Available in the market are several solutions for product design. All of them allow the user to design 3D models that will lead to the manufacture processes. Due to previous experience with the program and compatibility with the format supported by the 3D printer, the solution of choice is CATIA V5R19 by Dassault Systèmes, which delivers a combination of proven industry practices, knowledge and business processes. 3D design products and solutions such as CATIA V5R19 cover the entire shape design, styling and surfacing workflow. In the case of the Dassault Systèmes solution, the user is given real freedom to design any kind of complex shape thanks to easy to use shape design tools. Some of the most advanced functionalities are reverse engineering, Class-A surfacing, rapid propagation of design changes, powerful real-time diagnostic tools and highend visualization, making collaboration between design studios and engineering departments optimal. The user interface of CATIA V5R19, while intuitive, requires previous knowledge on software of the kind to be able to cope smoothly with the program, and specific formation on the program to completely benefit from the advantages of the solution. Fundació CIM, an attached entity to the Universitat Politècnica de Catalunya, as institutional objectives has transferring knowledge in engineering and technology management, as well as making tools available to professionals in order to bring academic and business realities closer (maximal technological competitiveness). Fundació CIM offers monthly courses on several 3D design software solutions, available to students, teaching staff and professionals equally. In July 2011, a complete course on CATIA V5R19 was offered and taken, acquiring therefore experience with the program, and making it the preferred 3D design tool of the trade. Due to the aforementioned compatibility with the 3D printer, it was selected to design the Lab-on- Spoon shape. Page 68 Report 5.1.1. CAD models to test the 3D printer MakerBot’s Replicator offers personalized manufacturing, as it allows transforming 3D designs into physical objects. It is a Desktop 3D Printer, meaning it is intended for single piece use, rather than multiple copies in bulk of the same model. This feature makes it a very good option for prototyping, as it allows the user to create master models replicable afterwards using machinery designed for mass production. This printer has been designed to be very user-friendly, with very intuitive interface and clear instructions in all the printing steps. Due to it not being excessively technical with said instructions, it does not require specialized workforce to assemble and put in operating conditions. With readily available 3D designs, it allows a quick start-up of the printing system. The available printer includes two extruders, with the option of dual printing. Due to the lack of familiarity with the device, as it is of recent purchase, before working with the final shape, some preliminary models are designed in order to test and calibrate the 3D printer. The first test model is the shape of an owl. The animal figure is simplified, while still maintaining the essence of its form. The model is almost solid, with some edges and rims to make the shape somehow complicated for the test print. It has rounded edges, in order to test the smoothness of the surface that can be reached with the 3D printer. Figure 5-1: Owl figure test print Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 69 The second model, aimed to test the possibility of assembling a figure parting from smaller snapping pieces, consists on the logotype of Technische Universität Kaiserslautern. It also allows the possibility of printing in different colours, one of the features of the printer. Figure 5-2: TU KL Logotype Mounted (web colours, non-print version) Figure 5-3: TU KL Logotype Pieces (actual colours, print version) Page 70 Report 5.1.2. CAD model for the concavity of Lab-on-Spoon The Lab-on-Spoon, as its name indicates, has its sensors and electronics mounted into a spoon shape, similar to the wooden spoons that have traditionally been used for cooking. A CAD model is designed to satisfy the needs of the Lab-on-Spoon while keeping a pleasant to the touch and the eye shape, as well as accomplishing some ergonomics objectives. Given the scope of the project, and knowing that in this first phase of the project the sensors and electrode will be connected to the Gecko Development Kit rather than a specially made PCB, the CAD model consists on the part of the concavity of the spoon-like shape. The handle is left aside, as it would result in a nuisance when manipulating the cables and connections to the Gecko Development Kit, although the concavity is designed with the prospect of having one attached in the near future. For this reason, a protuberance is added to the concavity, which can host the handle via a click-in system. In the near future, it is expected to design the handle so that an own PCB for the application can be placed inside. The concavity of the spoon is designed as similar as a cooking spoon as possible, taking the limitations of Catia V5R19 and the Desktop 3D Printer into account. The preliminary design is done with smooth surfaces and avoiding sharp edges, especially in those areas sensitive to be in contact with the user’s mouth. Considering the Desktop 3D Printer can have resolution problems when the shape is too intricate, and thus produce lumpy irregularities, the spoon is kept as simple as possible while still accomplishing its purpose (being able to hold liquids and stir thicker or denser substances). The design also takes into account the fact that the Desktop 3D Printer is unable to print without a layer underneath. With this is mind, some thin and small supports for those parts that would be aerial are added to the design. These supports are intended to be afterwards removed by sanding or cutting them, in order to achieve the final shape intended for the spoon. Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 71 In order to make the design process easier, a first draft of a full solid spoon is produced. Parting from it, a cavity to host the wiring of the Lab-on-Spoon connected to the sensors and electrode is introduced. Figure 5-4: Full solid spoon and full spoon with concavity (half view) CAD design Page 72 Report Considering the impossibility to print a full spoon with a cavity without having sacrificial printing material available, as well as an obvious difficulty to manipulate the wiring in such circumstance, the full spoon is split in two halves. The splitting plane is carefully chosen, considering the best option possible for the Desktop 3D Printer. Figure 5-5: Lab-on-Spoon halves CAD design Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 73 5.2. Preparation of CAD models for printing The Replicator needs to operate with GCode files (namely *.s3g), it cannot directly process the format of 3D design files. A slicing engine that prepares the models is needed. Available as freeware, the company offers MakerWare as the interface to transform 3D design files (it supports *.stl and *.obj) into the required format. This software includes a replica of the build platform of the Replicator, so that the user can add and place the desired objects to print, as well as scaling, moving or rotating individual models or groups of models at once. Due to The Replicator being an older model, files produced in MakerWare present some shortages. The extruder cannot be chosen, and is set on Left by default. The platform, although set to be heated, stays at room temperature during the print process. An older version of the available freeware, ReplicatorG, needs to be used to cover these lacks. Its interface is similar to that of the MakerWare, and as user friendly as this new freeware. However, as it is a more complex program, it covers the lacks of the new one. With the ReplicatorG, files necessary for printing are produced. As it allows the user to scale and rotate the model, said prints are created with the most obvious printing limitations in mind, so that the result is as smooth and easy to print as possible. The software also has a replica of the build platform, equipped with axes and divisions of the plate. The centre, the midpoint of the length and width, and the overall height of the printing area are clearly delimitated, with stronger lines than the divisions of the plate. This way, placing the desired object to print becomes easy, giving a detailed perspective of what the outcome will be. As stated before, the available Desktop 3D printer is a dual extruder model. Most of the files are created for both of them, in an attempt to optimize the usage of the extruding plastic rolls and have a second printing possibility if problems with the first were encountered. Catia V5R19 works with its own format of files (*.CATPart), but allows the user to export the created 3D designs by saving them into *.stl format, which is supported by both the MakerWare and the ReplicatorG. Once the 3D design is saved in the suitable format, the user can start working with the slicing engine tool of choose (in this case, ReplicatorG). The “Machine” drop-down menu relates to the kind of 3D printer that the created files are used with. Under “Machine Type (driver)”, the selected option must be The Replicator Dual. No other options require changing, as the created files are transferred to the printer using an SD-card, a much simpler version than the serial connection. Page 80 Report 6. Prototype The final version of the concavity of the spoon is printed in two different sizes, a small scale one for experimentation purposes, and a full scale model. These pieces are left for future continuators of the Lab-on-Spoon project. The prototype in this stage of the project is a very rough version of the connectivity of the sensors to the microprocessor, intended for testing and debugging purposes. It allows freedom to add and remove elements from the circuit, as well as easy manipulation of said elements. As it is a flexible design, most of the geometry of the circuit can be moved to accommodate the necessities of the project. In the case a replacement is needed, it is possible to change the broken element quickly, as it is comfortably reachable for the user. The design reached in this phase will be used in the future to accommodate the Lab-on- Spoon on a dedicated PCB. The electronics are mounted on a blank BE 439 one-side prototyping board. Connectors are installed in the power and ground lines, as well as in the I2C communication lines. They are prepared to be properly attached to the EFM32-G890-F128 Gecko Development Kit, where the microprocessor is currently located. It also enables the possibility of better transportation and storage of the prototype, as the most delicate parts can be separated and protected adequately. The AD5933 has a default configuration meant for sensing impedances between 1 kΩ and 10 MΩ, and an extended configuration for sensing impedances below 1 kΩ. The differences between both configurations allow them to be mounted on the same board, requiring minor soldering to enable one or the other. Figure 6-1: Configuration A – Default – Range 1 kΩ to 10 MΩ Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 81 On this stage of the Lab-on-Spoon project, only Configuration A (Figure 6-1) is mounted, for the sake of performing tests on known electric circuits to ensure the proper functioning of the AD5933 chip, as well as the feasibility of the data collected. The configuration is replicated on two prototyping boards. Figure 6-2: Configuration B – Additional circuitry – Range 100 Ω to 1 kΩ Figure 6-3: Prototype A Page 82 Report An important component of both configurations is the current-to-voltage amplifier gain resistor or feedback resistor, noted as RFB in Figure 6-1 and Figure 6-2. It is dependent on the value of the unknown impedance, and must be set knowing the range of variation of said impedance. The calculation of RFB can be done by solving the equations for non-saturation of the internal operational amplifier of the AD5933. 2 222 2 0 VDD V V VDD Z RFB RFB V VDD Z VDD V VDD VV II s o unknown o unknown s           (Eq. 6.1) Figure 6-4: Prototype B Figure 6-5: Circuit for RFB calculation Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 83 By properly substituting the values of VDD and Vs, and knowing the maximum and minimum values of Vo correspond to those the ADC is able to process, a range for the relation between the current-to-voltage amplifier gain resistor and the unknown impedance can be established and, thus, a proper value for RFB selected. This equation must consider the PGA setting of the AD5933 chip, which can be set to 1 or 5. When RFB is not properly selected taking PGA gain into account, the ADC can saturate. Lines SCL and SDA, the I2C bus communication lines, relate to pins 14 and 15 on port D of the microprocessor. They are routed on location 3, available to the user through pins 15 and 16 on port J of the prototype board on the EFM32-G890-F128 Gecko Development Kit. Power at 3,3 V and ground are also accessible on the prototype board. Figure 6-6: Connections on the PB of the EFM32-G890-F128 Gecko Development Kit Page 84 Report 7. Experiments in application 7.1. Calibration of the system 7.1.1. System phase calculation The AD5933 has its own system phase, as the internal components of the system and the fact that it is not ideal contribute to alter the phase and add a certain value to it. When calculating the phase of the desired sample, it must be considered that it is, in fact, the difference between the phase of the system with the sample connected and the system on its own. The phase measured accounts for the phase shift introduced to the DDS output signal as it passes through the internal amplifiers on the transmission and receive side of the AD5933 along with the low-pass filter and also the sampled impedance. The system phase is, thus, a calibration value necessary to obtain the true value of the phase. systemsystemsamplesample    (Eq. 7.1) With the purpose of not introducing additional phase lead or lag to the AD5933 signal path, a resistor is placed across said pins as the calibration impedance (Zcalibration). A frequency sweep is run and, with the values of the real and imaginary registers, the phase angle is calculated in radian using the arctangent formula and then transformed to degrees, as described in Table 7-1 and depending on the quadrant the measured impedance belongs to. Real Imaginary Quadrant Phase Angle + + 1st  º R I tan 180 1        - + 2nd                 º R I tanº 180 180 1 - - 3rd                 º R I tanº 180 180 1 + - 4th                 º R I tanº 180 360 1 Table 7-1: Phase angle calculation in degrees Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 85 7.1.2. Gain factor calculation Due to the system not being ideal, a calibration term or scaling factor must be taken into account when calculating the DFT magnitude of the desired sample. 22 IRMagnitude  (Eq. 7.2) The gain factor is obtained through the use of a known calibration impedance (Zcalibration) connected between the output and input pins and measuring the resulting magnitude of the code. By the use of a resistor as the calibration impedance, the whole calibration process can be performed in a single experiment, calculating both system phase and gain factor using the same data collection. 22 1 1 IR Z Magnitude mpedanceI Code cetanAdmit FactorGain ncalibratio                           (Eq. 7.3) The gain factor shows a variation with frequency due to the finite frequency response of the AD5933, resulting in an error in the impedance calculation over a frequency range. To minimize this error, the frequency sweep is limited to as small a frequency range as possible. This calibration term must be recalculated for changes in the current-to-voltage gain setting resistor (RFB), the output excitation voltage and the PGA gain. To measure the impedance of unknown impedances (food samples in the case of the Lab-on-Spoon), the gain factor is used as follows. 22 11 IRFactorGain MagnitudeFactorGain mpedanceI     (Eq. 7.4) Page 86 Report 7.1.3. Experimental determination of calibration parameters To calculate the gain factor and the system phase, a calibration resistor of nominal value 220 kΩ is used. The datasheet for the AD5933 suggests a calibration resistor of 200 kΩ, the one used in the experiment is the closest available. In order to obtain these parameters with the highest grade of accuracy as possible, the actual value of the resistor is measured with a multimeter, obtaining 219 kΩ. The current-to-voltage amplifier gain resistor is set to a nominal value of 220 kΩ, again following recommended values for this experiment found on the datasheet for the AD5933, which state both resistors to be the same. By the use of suggested resistor values, illustrative register readings are available through the datasheet for the AD5933, as it includes typical values for real and imaginary data under said calibration conditions. The experiment is conducted using the same parameters that are used for further experimentation, described in Section 4.1.3. Hexadecimal Twos Complement Decimal Frequency Re FFFF 1111 1111 1111 1111 -1 30 kHz Im 040B 0000 0100 0000 1011 1035 Re FFFF 1111 1111 1111 1111 -1 31 kHz Im 040F 0000 0100 0000 1111 1039 Re FFFF 1111 1111 1111 1111 -1 32 kHz Im 0412 0000 0100 0001 0010 1042 Re FFFF 1111 1111 1111 1111 -1 33 kHz Im 0415 0000 0100 0001 0101 1045 Re FFFF 1111 1111 1111 1111 -1 34 kHz Im 0414 0000 0100 0001 0100 1044 Re FFFF 1111 1111 1111 1111 -1 35 kHz Im 041D 0000 0100 0001 1101 1053 Re FFFF 1111 1111 1111 1111 -1 36 kHz Im 0422 0000 0100 0010 0010 1058 Re FFFF 1111 1111 1111 1111 -1 37 kHz Im 0423 0000 0100 0010 0011 1059 Re FFFF 1111 1111 1111 1111 -1 38 kHz Im 0422 0000 0100 0010 0010 1058 Re FFFF 1111 1111 1111 1111 -1 39 kHz Im 0428 0000 0100 0010 1000 1064 Table 7-2: Real and Imaginary data for calibration experiment Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 87 Using the formulas described in the previous sections, the gain factor and phase angle are calculated for each frequency. For the purpose of simplifying further experimentation, the calibration parameters are set as the mean value of the parameter for each frequency. The gain factor is a dimensionless parameter; the phase angle is expressed in degrees. Frequency Magnitude Gain Factor Phase Angle 30 kHz 1035,00048 4,4118E-09 90,05535822 31 kHz 1039,00048 4,39481E-09 90,0551451 32 kHz 1042,00048 4,38216E-09 90,05498634 33 kHz 1045,00048 4,36958E-09 90,05482848 34 kHz 1044,00048 4,37376E-09 90,054881 35 kHz 1053,00047 4,33638E-09 90,05441193 36 kHz 1058,00047 4,31589E-09 90,05415478 37 kHz 1059,00047 4,31181E-09 90,05410365 38 kHz 1058,00047 4,31589E-09 90,05415478 39 kHz 1064,00047 4,29155E-09 90,0538494 4,35019E-09 90,05458737 As it can be observed in Table 7-3, it is sensible to use the mean value of the gain factor and the phase angle, for the variation with frequency is small enough. This is, however, due to the fact that a small range of frequencies has been used in the test, for the sake of minimizing the error that comes from the variation of the gain factor with frequency. It must be taken into account if the frequency sweep was to be altered, as it might not be reasonable to use the mean value in other cases. At this point, a possible anomaly in the performance is detected, as the real register should present a constant, positive value in a larger scale. According to the datasheet, for a 200 kΩ configuration, the typical values of the spectra are 0x227E for the imaginary data register and 0xF064 for the real data register. When the impedance of a resistor is increased, the imaginary data registers decreasing values, thus making the reading on the experiment feasible. On the contrary, the real data register’s only concordance is the sign of the value. Table 7-3: Gain Factor and Phase Angle calculation Page 88 Report 7.2. Experiments in application with electronic networks Experiments are performed with known electronic networks, to test the feasibility of the data collected by the AD5933 and a proper calibration. They are performed on Prototype B (Figure 6-4) using the source code described in the Appendices. The first experiment subjects a single 100 kΩ resistor to the previously set test conditions, and uses the gain factor and system phase to calculate the impedance and phase of the sample. Hexadecimal Twos Complement Decimal Frequency Re FFFF 1111 1111 1111 1111 -1 30 kHz Im 084D 0000 1000 0100 1101 2125 Re FFFF 1111 1111 1111 1111 -1 31 kHz Im 084D 0000 1000 0100 1101 2125 Re FFFF 1111 1111 1111 1111 -1 32 kHz Im 084B 0000 1000 0100 1011 2123 Re FFFF 1111 1111 1111 1111 -1 33 kHz Im 0845 0000 1000 0100 0101 2117 Re FFFF 1111 1111 1111 1111 -1 34 kHz Im 0842 0000 1000 0100 0010 2114 Re FFFF 1111 1111 1111 1111 -1 35 kHz Im 0845 0000 1000 0100 0101 2117 Re FFFF 1111 1111 1111 1111 -1 36 kHz Im 0842 0000 1000 0100 0010 2114 Re FFFF 1111 1111 1111 1111 -1 37 kHz Im 0840 0000 1000 0100 0000 2112 Re FFFF 1111 1111 1111 1111 -1 38 kHz Im 0838 0000 1000 0011 1000 2104 Re FFFF 1111 1111 1111 1111 -1 39 kHz Im 083A 0000 1000 0011 1010 2106 Table 7-4: Real and Imaginary data for 100 kΩ resistor experiment Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 89 As observable in Figure 7-1, the value of the impedance is higher than expected, possibly due to the system not being ideal. A slight variation with frequency is also noticed. The Nyquist plot of the experiment shows a straight line parallel to the real axis, the expected behaviour for a single resistor circuit. The covered region of the polar representation is very small, as the frequency sweep is done on a narrow range of frequencies. Figure 7-1: Impedance profile for 100 kΩ resistor experiment Figure 7-2: Nyquist plot for 100 kΩ resistor experiment Page 96 Report 7.4. Problems during experiments in application When constructing the prototypes to be used during the experiments in application, with the purpose of ensuring the tests to be performed, two boards were mounted. Should one of them have a problem, the other would be ready for use and continue the experimentation. In order to test their good functioning in a short time, the source code was altered so that only one register was written to and read from. Said code made use of the AD5933_RegisterSet and AD5933_RegisterGet functions, fully described in Appendix C: i2c_AD5933.c. The least significative bits of the Start Frequency (register 0x87) are programmed and the reading of the register is displayed in the LCD screen of the EFM32-G890-F128 Gecko Development Kit. This shortened version of the source code was downloaded onto the microprocessor and tested with both prototypes, with the purpose of ensuring I2C functionality and/or communication. The first board that was used (Figure 6-3: Prototype A) was able to communicate with the microprocessor intermittently. Upon reset of the microprocessor, after a successful communication with the AD5933 chip on an immediately previous connection, an error of negative acknowledgement (NACK) was received during transfer, meaning that the slave (in this case, the AD5933 chip) was not in the I2C bus. Due to the intermittency of the ability to communicate and the fact that the slave device was physically connected to the bus, a problem during soldering was an educated guess, most likely due to a cold joint. The soldering was reviewed and corrected or redone on possibly conflictive joints. However, the intermittent NACK error kept appearing when trying to communicate, until it became permanent. The board was left for further inspection under the supervision of better trained eyes, as the hypothetical source of the problem could not be found. The second board (Figure 6-4: Prototype B) proved to have no communication problem after the single-register test was performed multiple times and, thus, the entirety of the source code was downloaded onto the microprocessor. In the process of performing a frequency sweep, the Status Register (Register Address 0X8F) did not present valid real and/or imaginary data, meaning that the frequency sweep was not performed. Upon further investigation of the possible reasons of this lack of data, it was found that, in an attempt to maximize the speed the frequency sweep was performed at, the amount of settling time between programming the command “Initialize with Start Frequency” and the command “Start Frequency Sweep” to the Control Register had not been respected. Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 97 A delay between said processes was added, fixing the presence of valid data and, thus, enabling the performance of a frequency sweep. The need of such a delay was documented and reviewed on Section 3.3.4. At this point, the first experiments were run and monitored, encountering the facts described in Section 7.3. On the process of modifying the analogue front end of the second board (Figure 6-4: Prototype B), the behaviour of the AD5933 became erratic upon repetition of the experiments under the same conditions. Although the Control Register had been properly programmed, the output sine wave in one of the last experiments presented an output voltage range of 1 V peak-to-peak, instead of the programmed and therefore expected 2 V. On the next power-up of the system, the error regarding the unavailability of valid real and/or imaginary data was present again. The solution previously described was put into practice, increasing the aforementioned delay between programming the “Initialize with Start Frequency” command and the ”Start Frequency Sweep” command. The problem remained unsolved. Predicting a possible over-performance of the AD5933 chip, it was left to rest for 30 minutes before starting the system up again, with unsuccessful results. After one day, the chip was put to test again, with no success on performing the sweep again. At this point, it was handed to David Los Arcos, continuator of the Lab-on-Spoon project. Recent reports state that after approximately five days after the last start-up the AD5933 chip did no longer present this error regarding the validity of the data, being able to perform frequency sweeps again. 7.5. State after experiments in application David Los Arcos, continuator of the Lab-on-Spoon project, received the two constructed prototypes with the AD5933 chip mounted on them, as well as the necessary pins and connectors to feed the boards. Prototype A, as stated before, registered a NACK error during any I2C communication sequence, possibly due to a cold joint. The board was handed to him with instructions about the need of revision of the soldering by more experienced eyes. Prototype B was prepared for its handing after the last unsuccessful start-up described in the previous section. The board was presented to him with the default configuration described in Figure 6-1 and the electronic components used to expand the circuit soldered on the board. The wiring followed the simplest configuration (Figure 6-1); with the purpose of ensuring that the invalidity of the data was not produced by the extra circuitry. Page 98 Report The default configuration, albeit returning faulty values, had been proved functional and was recovered for the sake of testing the good functioning of the AD5933. The board also included small known electronic networks to be used as the unknown impedance when performing frequency sweeps. Recent reports state a recovery of the board after a five day time lapse from the last unsuccessful start-up. David Los Arcos was updated on the problems experienced after performing the recorded tests and received documentation regarding possible solutions to both the invalidity of the data and the saturation of the ADC. He was informed of the need of a delay in the programming in order to be able to obtain valid data. The untested possibilities to prevent the ADC from saturating include the use of the extended circuit not on VOUT, but after the signal has gone through the unknown impedance, as the objective is to maintain the DC value of the wave centred on VDD/2 when this enters the AD5933 chip for processing. Another presented solution is the use of said circuitry on both ends of the unknown impedance. He is also recommended to use operational amplifiers whose single feed can be set to 3,3 V, so that it is the same as VDD and he does not encounter problems derived from this fact. The handed prototype is half-functional, as it performs what it is expected of it but returns invalid results. The programming of the AD5933 is done flawlessly and swiftly, proving the good nature of the source code. Consequently, the frequency sweep is carried out accordingly to selected programming values. This can be seen through an oscilloscope, with the probe located on VOUT pin: in the expected lapse of time, the frequency increases and thus the period of the wave decreases. It is observed that the sine wave becomes narrower with every step. Regardless of their validity, real and imaginary data are obtained, meaning the DFT operation is properly performed and the impedance is adequately extracted from the faulty values. These troublesome results do not allow suitable observation and experimentation of samples, but encourage the extension of the analogue circuitry in order to condition the signal before its processing. Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 99 8. Conclusions The first stage of the Lab-on-Spoon project consists on an exhaustive study of the sensor possibilities for implantation in an ambient assisted living device for smart kitchen scenarios, seeking to intertwine sensor technology with cooking, a traditional and slightly oldfashioned field. One of the biggest goals of the project is to achieve a fully functioning Lab-on- Spoon with minimal power consumption. For this purpose, the family of energy-friendly microcontrollers included in Energy Micro’s Gecko Development Kits are used in this application. Given the time budget, after said sensor study it is decided that the focus of the first stage of the project is electrochemical impedance spectroscopy. The concept of impedance spectroscopy is fatherly examined and given context of a smart spoon appliance, considering it suitable for the sensing of the salinity degree of a food sample. An implementation of a frequency sweep apt for impedance spectroscopy is achieved by the use of an integrated chip, AD5933, the functionalities of which allow the user to program registers related to the frequency range of the sweep, as well as other parameters directly dependant on the sample being subjected to test. For the programming of the AD5933 chip, I2C communication is studied, as the information is transferred between the microprocessor and the chip via the aforementioned protocol. A code is developed from scratch to fulfil said communication. Some primitive tests were performed in order to check the good functioning of the AD5933 and ensure proper communication with the microprocessor. While they proved the communication efficiency, it was made obvious the analogue front end of the sample needed to be enlarged, with the purpose of further preparing the signal from the sample for processing, as well as opening a discussion regarding the frequency range of the sweep performed on the sample or, in other words, the range of the impedance spectroscopy. It is concluded that the implementation of sensor technology in the context of a smart kitchen scenario is not only useful but innovative, as it is a field where old-fashioned ways prevail. The use of impedance spectroscopy allows exhaustive analysis of unknown food samples, which can be categorized by comparison. On the same line, it is possible to determine the proper realization of a dish by comparing the data collected with that of an electronic recipe book. The selected chip for performing impedance spectroscopy sweeps needs further study, as it has its limitations and requires to be adapted in order to properly use it in the Lab-on-Spoon application. Page 100 Report 9. Appreciations To my parents, Montserrat Miralles and Juan Manuel Espejo, for all you have given me in my life. For always being there when I most needed counselling, for teaching me the true value of things, for the education you have given me, for your constant support and encouragement through my schooling years, for giving me the opportunity to travel abroad to complete my studies (albeit the sacrifice this has been for all of us). Thank you for your constant motivation and for raising me to be the person I am now. But above all, thank you for putting up with me, for being proud of me, for your love and dedication. I owe you everything. To my colleagues Jaime Rodríguez and Manel Escudero, who I can proudly call dear friends, for the uncountable smiles you have drawn on my face. We went through hell and beyond, but still survived, probably because we were together. Our “electronic bond” is priceless to me. Thank you for your craziness, for cheering me up, for your expertise when I ran out of ideas, for making University worth it. To my professors in Barcelona, Josep Rius and Manuel Moreno, who have always inspired me and fed my passion for Electronics. You are the engineers I look up to, not only because I consider you wise beyond measure, but also for your friendly predisposition to help your students. Your lessons have had a great impact in me, and for that I am thankful. To David Los Arcos, my laboratory partner, and Nicu Muntean, for helping me out in the last stages of this project, both personally and professionally. I can only wish you both the best of lucks. To my supervisor, Prof. Dr.-Ing. Andreas König, for the suggestion of the topic and giving me the opportunity to work on it, and for his guidance and counselling during the months this project has lasted. Thank you, from the bottom of my heart. Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 101 10. Bibliography 10.1. Bibliographic references [1] MASSACHUSSETTS INSTITUTE OF TECHNOLOGY MEDIA LAB. COUNTER INTELLIGENCE GROUP. Intelligent Spoon. Massachussetts, 2006. [http://www.media.mit.edu/ci/projects/intelligentspoon.html, 5th of May of 2012]. [2] WINDOWS TO THE UNIVERSE TEAM. NATIONAL EARTH SCIENCE TEACHERS ASSOCIATION. CTD Instrument. Colorado, 2001. [http://www.windows2universe.org/earth/Water/CTD.html&lang=en, 3rd of July of 2012] [3] INCZÉDY, J., LENGYEL, T. and URE, A. M. International Union of Pure and Applied Chemistry. Compendium of Analytical Nomenclature, Definitive Rules. Oxford: Blackwell Science, 1998, “8.3.2.4 Ion-selective field effect transistor (ISFET) devices”. [4] EDITUM. LAMBERZ, S. Cocina: Temperaturas Seguras, Alimentos Confiables. Afganistan, 2008. [http://www.editum.org/Cocina-Temperaturas-Seguras-Alimentos-Confiables-p-997.html, 7th of May of 2012] [5] GASTRONOMÍA & CÍA. Métodos de Cocción. España, 2005-2012. [http://www.gastronomiaycia.com/tag/metodos-de-coccion/, 3rd of June 2012] [6] DULCES DE QUECA. GENERAL. Tabla de temperaturas del horno. Peru, 2007. [http://www.dulcesdequeca.com/general/tabla-de-temperaturas-del-horno.html, 26th of June 2012] [7] OMEGA. PRODUCT INFORMATION. Thermocouples. Omega Engineering Technical Reference, 2003. [http://www.omega.com/prodinfo/thermocouples.html, 1st of August 2012] [8] CHAMBERS, R. G. Thermoelectric effects and contact potentials. Physics Education. Issue 6, September 1977, p. 374-380. Page 102 Report [9] OMEGA. PRODUCT INFORMATION. Thermistor. Omega Engineering Technical Reference, 2003. [http://www.omega.com/prodinfo/thermistor.html, 1st of August 2012] [10] TEMPERATURES.COM. SENSORS. Resistance Temperature Detectors. Temperature.com Inc, 1999. [http://www.temperatures.com/rtds.html, 1st of August 2012] [11] CHIRAS, D. D. Human Biology. Sudbury: Jones and Bartlett Publishers, 2005, p. 201- 202. [12] MEYERHOF, W., BEHRENS, M., BROCKHOFF, A., BUFE, B. and KUHN, C. Human Bitter Taste Perception. Chemical Senses. Issue 30, Supplement 1, 2005, p. i14-i15. [13] COVINGTON, A. K., BATES, R. G. and DURST, R. A. Definition of pH scales, standard reference values, measurement of pH and related terminology. Pure & Applied Chemistry. Vol. 57 No. 3, 1985, p. 531-542. [14] INSTITUTO NACIONAL DE TECNOLOGÍA INDUSTRIAL DE ARGENTINA. COOPERACIÓN IBEROAMERICANA DE CIENCIA Y TECNOLOGÍA PARA EL DESARROLLO. Sensor de pH tipo ISFET. Argentina, 2001. [http://www.inti.gov.ar/citei/cyted/isfet.htm, 7th of May of 2012] [15] HABARA, M. and TOKO, K. Discrimination of Saltiness with Coexisting Components using Multichannel Taste Sensor with Lipid Membranes. IEICE transactions on electronics. E. 83, C. 7, 2000, p. 1040-1045. [16] SUITE101. FERNÁNDEZ, Á. La sal, sus tipos, usos, beneficios y contraindicaciones. Spain, 2010. [http://suite101.net/article/la-sal-tipos-usos-beneficios-y-contraindicaciones-a20288,, 10th of July of 2012] [17] LIVESTRONG.COM. RICE, L. 5 things you need to know about recommended daily sodium intake. California, 2011. [http://www.livestrong.com/article/4734-need-recommended-daily-sodium-intake/, 12th of July of 2012] [18] DA ROCHA, Rogério T., GUTZ, Ivano G.R. and DO LAGO, Claudimir L., A Low-Cost and High-Performance Conductivity Meter. Journal of Chemical Education. Vol. 74 No. 5, 1997, p. 572-574. Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 103 [19] TILLEY, R. J. D. Understanding solids: the science of materials. West Sussex: John Wiley & Sons, 2004, p. 280-282. [20] RAMOS, P., DIAS, J. M., GEIRINHAS, H. M. and LOPES, A. A Four-Terminal Water- Quality-Monitoring Conductivity Sensor. IEEE Transactions on Instrumentation and Measurement. Vol. 57 No. 3, 2008, p. 577-582. [21] GAMRY INSTRUMENTS, Inc., Basics of Electrochemical Impedance Spectroscopy (Application Note). Pennsylvania: 2010. [22] GAMRY INSTRUMENTS, Inc., A Snapshot of Electrochemical Impedance Spectroscopy (Application Note). Pennsylvania: 2011. [23] CHINAGLIA, D.L., GOZZI, G., ALFARO, R.A.M. and HESSEL, R. Espectroscopia de impedância no laboratório de ensino. São Paulo: 2009. [24] CHANG, Byoung-Yong and PARK, Su-Moon, Electrochemical Impedance Spectroscopy. Annual Review of Analytical Chemistry. Vol. 2010.3, 2010, p. 207-229. [25] OWEN, D. Glossary: Quantization. Wavelenght Media, 1995-2008. [http://www.mediacollege.com/glossary/q/quantization.html, 15th of October of 2012] [26] SHANNON, Claude E., Communication in the presence of noise. Proceedings Institute of Radio Engineers. Vol. 1.37, 1949, p. 10-21. [27] RHEA, R. W. Oscillator Design & Computer Simulation. Unitated States of America: McGraw-Hill, 1996, p. 111. [28] RUNDLE, Chris C., A Beginners Guide to Ion-Selective Electrode Measurements. London, 2000-2012. [http://www.nico2000.net/Book/Guide1.html, 3rd of January of 2013] [29] OMEGA. PRODUCT INFORMATION. High Temperature Insertion Type Electrode. Omega Engineering Technical Reference, 2005. [http://www.omega.com/pptst/PHE543110.html, 3rd of January of 2013] Page 104 Report 10.2. Complementary bibliography [A] CHEN, M. C. The Design of Intelligent Cookware. Massachusetts: 2003. [B] LAKE SHORE CRYOTRONICS, INC. Temperature/Resistance Table for Platinum Sensors. DIN IEC 751. Form Number F038-00-00 Revision 0. Westerville: Lake Shore Cryotronics, 2000. [C] INNOVATIVE SENSOR TECHNOLOGY. Platinum-Temperature Sensors. Wattwill: Innovative Sensor Technology, 2009. [D] INNOVATIVE SENSOR TECHNOLOGY. Platinum – 600ºC. Platinum Thin-Film Temperature Sensor 2,3x2mm. Wattwill: Innovative Sensor Technology, 2006. [E] INNOVATIVE SENSOR TECHNOLOGY. Platinum – 600ºC MiniSens. The World’s Smallest Platinum Thin-Film Temperature Sensor. Wattwill: Innovative Sensor Technology, 2006. [F] LABFACILITY. Pt100 Platinum Sensing Resistors. West Sussex: Labfacility, 2008. [G] LABFACILITY. Pt1000 Platinum Sensing Resistors. West Sussex: Labfacility, 2008. [H] AKIYAMA, T., NIKI, E. Ion-sensitive field-effect transistor for pK and pNa sensing. Pure & Applied Chemistry. Vol. 59 No. 4, 1987, p. 535-538. [I] WAGNER, Mauricio O. Acidez y Ph. Argentina: Cerveceros Caseros, 2005. [J] TROXLER, S., REARDON, J. W. North Carolina Department of Agriculture and Consumer Services. Food and Drug Protection Division. pH y los Alimentos. North Carolina: 2012. [K] BUDDING, R.W., STRACKEE, L. A dynamic admittance meter base on a voltage controlled oscillator IC. Review of Scientific Instruments. Vol. 49 No. 2, 1975, p. 210- 212. [L] ALDOSKY, Haval Y. Yacoob, SHAMDEEN, Suzan M. H. A new system for measuring electrical conductivity of water as a function of admittance. Journal of Electrical Bioimpedance. Vol. 2, 2011, p. 86-92. [M] ZHANG, Yi. A Design of Complex Impedance Meter. Ithaca: 2007. Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 105 [N] GONZÁLEZ, César Antonio. Validación de la terapia guiada por espectroscopía de impedancia eléctrica gástrica en un modelo experimental de choque séptico inducido. México D.F.: 2007. [O] LASIA, A. Electrochemical Impedance Spectroscopy and Its Applications, Modern Aspects of Electrochemistry. Kluwer Academic/Plenum Publishers. New York, 1999, Vol. 32, p. 143-248. [P] OTHMAN, Salah. Espectrómetro de Impedancia para Monitoreo de Daño Isquémico Tisular. México D.F.: 1999. [Q] OTHMAN, Salah, SACRISTÁN, Emilio. Espectrómetro de Impedancia Compleja para Aplicaciones Biomédicas. Sociedad Cubana de Bioingeniería. Artículo 00367. La Habana: 2001. [R] GONZÁLEZ, César A., SACRISTÁN, Emilio, VILLANUEVA, Cleva, OTHMAN, Salah, NARVÁEZ, Raúl, ALJAMA, Tomás. Espectroscopía de Impedancia para Monitoreo de Daño Isquémico en la Mucosa Intestinal. Sociedad Cubana de Bioingeniería. Artículo 00444. La Habana: 2001. [S] MACDONALD, J. Ross. Impedance Spectroscopy. Annals of Biomedical Engineering. Vol. 20, 1992, p. 289-305. [T] BARSOUKOV, Evgenij, MACDONALD, J. Ross. Impedance Spectroscopy. Theory, Experiment and Applications. Hoboken: John Wiley & Sons, 2005. [U] ANALOG DEVICES, INC. AD5933 – 1 MSPS, 12-bit Impedance Converter, Network Analyzer – Datasheet. Norwood: Analog Devices, Inc., 2011. [V] ANALOG DEVICES, INC. Evaluating the AD5933 1 MSPS, 12-bit Impedance Converter, Network Analyzer – User Guide 364. Norwood: Analog Devices, Inc., 2012. [W] ENERGY MICRO. Cortex-M3 Reference Manual. Oslo: Energy Micro, 2011. [X] ENERGY MICRO. EFM32-G890 Datasheet. Oslo: Energy Micro, 2012. [Y] ENERGY MICRO. EFM32 G890 MCU Board. Oslo: Energy Micro, 2010. [Z] ENERGY MICRO. Quick Start Guide – EFM32 Gecko Development Kit. Oslo: Energy Micro, 2010. Page 112 Report /**************************************************************************//** * @brief Delays number of msTick Systicks (typically 1 ms) * @param dlyTicks Number of ticks to delay *****************************************************************************/ void Delay(uint32_t dlyTicks) { uint32_t curTicks; curTicks = msTicks; while ((msTicks - curTicks) < dlyTicks) ; } /**************************************************************************//** * @brief Initialization of LCD and LEDs for debugging when connected to DVK. *****************************************************************************/ void LCD_LED_Initialize(void) { /* Initialize LCD controller without boost */ SegmentLCD_Init(false); SegmentLCD_Write("HELLO"); /* Initialize DK board register access */ BSP_Init(BSP_INIT_DEFAULT); /* If first word of user data page is non-zero, enable eA Profiler trace */ BSP_TraceProfilerSetup(); /* "Clean" LEDs */ BSP_LedsSet(0x0000); } /**************************************************************************//** * @brief Turn off LCD and LEDs for debugging when connected to DVK. *****************************************************************************/ void LCD_LED_TurnOff(void) { /* "Clean" LEDs */ BSP_LedsSet(0x0000); /* Etiquette for turn-off and "clean" LCD */ SegmentLCD_Init(false); SegmentLCD_Write("BYE-BYE"); Delay(3000); SegmentLCD_Init(false); } Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 113 /***************************************************************************//** * @brief * Initalize basic I2C master mode driver for use on the DVK. * * @details * This driver only supports master mode, single bus-master. In addition * to configuring the EFM32 I2C peripheral module, it also configures DVK * specific setup in order to use the I2C bus. * * @param[in] init * Pointer to I2C initialization structure. ******************************************************************************/ void I2C_AD5933_Init(void) { int i; /* Initialize I2C driver for the AD5933 sensor, */ /* connected to the DK, using standard rate. */ /* Devices on DK itself supports fast mode, */ /* but in case some slower devices are added on */ /* prototype board, we use standard mode. */ I2C_Init_TypeDef i2cInit = I2C_INIT_DEFAULT; /* Initialize DVK board register access */ BSP_Init(BSP_INIT_DEFAULT); BSP_PeripheralAccess(BSP_I2C, true); CMU_ClockEnable(cmuClock_HFPER, true); CMU_ClockEnable(cmuClock_I2C0, true); /* Use location 3: SDA - Pin D14, SCL - Pin D15 */ /* Output value must be set to 1 to not drive lines low... */ /* We set SCL first, to ensure it is high before changing SDA. */ GPIO_PinModeSet(gpioPortD, 15, gpioModeWiredAnd, 1); GPIO_PinModeSet(gpioPortD, 14, gpioModeWiredAnd, 1); /* In some situations (after a reset during an I2C transfer), the slave */ /* device may be left in an unknown state. Send 9 clock pulses just in case. */ for (i = 0; i < 9; i++) { /* * TBD: Seems to be clocking at appr 80kHz-120kHz depending on compiler * optimization when running at 14MHz. A bit high for standard mode devices, * but DVK only has fast mode devices. Need however to add some time * measurement in order to not be dependable on frequency and code executed. */ GPIO_PinModeSet(gpioPortD, 15, gpioModeWiredAnd, 0); GPIO_PinModeSet(gpioPortD, 15, gpioModeWiredAnd, 1); } /* Enable pins at location 3 (which is used on the DVK) */ I2C0->ROUTE = I2C_ROUTE_SDAPEN | I2C_ROUTE_SCLPEN | (3 << _I2C_ROUTE_LOCATION_SHIFT); I2C_Init(I2C0, &i2cInit); /* Clear and enable interrupt from I2C module */ NVIC_ClearPendingIRQ(I2C0_IRQn); NVIC_EnableIRQ(I2C0_IRQn); } Page 114 Report /***************************************************************************//** * @brief * Program a piece of data into a given register * * @details * The user gives a piece of data (reg_write_data) to be written on a * certain register (reg_address) of a peripheral device (addr). * * @param[in] i2c * Pointer to I2C peripheral register block. * * @param[in] addr * I2C address for AD5933 sensor, in 8 bit format, where LSB is reserved * for R/W bit. * * @param[in] reg_address * Register to write. * * @param[out] reg_write_data * Data to write in the register. * * @return * Returns 0 if register written, <0 if unable to write register. ******************************************************************************/ int AD5933_RegisterSet(I2C_TypeDef *i2c, uint8_t addr, uint8_t reg_address, uint8_t reg_write_data) { I2C_TransferSeq_TypeDef seq; uint8_t data[2]; seq.addr = addr; seq.flags = I2C_FLAG_WRITE; /* Set register to be written and data to be transferred*/ data[0] = reg_address; seq.buf[0].data = data; data[1] = reg_write_data; seq.buf[0].len = 2; /* Do a polled transfer */ I2C_Status = I2C_TransferInit(i2c, &seq); while (I2C_Status == i2cTransferInProgress) { /* Enter EM1 while waiting for I2C interrupt */ EMU_EnterEM1(); /* Could do a timeout function here. */ } return(I2C_Status); } Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 115 /***************************************************************************//** * @brief * Read sensor register content. * * @details * The user reads a piece of data on a * certain register (reg) of a peripheral device (addr). * * @param[in] i2c * Pointer to I2C peripheral register block. * * @param[in] addr * I2C address for AD5933 sensor, in 8 bit format, where LSB is reserved * for R/W bit. * * @param[in] reg * Register to read. * * @param[out] val * Reference to place register read. * * @return * Returns 0 if register read, <0 if unable to read register. ******************************************************************************/ int AD5933_RegisterGet(I2C_TypeDef *i2c, uint8_t addr, uint8_t reg_address, uint8_t *val) { I2C_TransferSeq_TypeDef seq; uint8_t data[1]; seq.addr = addr; seq.flags = I2C_FLAG_WRITE; /* Select register to write */ seq.buf[0].data = ®_address; seq.buf[0].len = 1; /* Do a polled transfer */ I2C_Status = I2C_TransferInit(I2C0, &seq); while (I2C_Status == i2cTransferInProgress) { /* Enter EM1 while waiting for I2C interrupt */ EMU_EnterEM1(); /* Could do a timeout function here. */ } seq.flags = I2C_FLAG_WRITE_READ; /* Select register to be read */ seq.buf[0].len = 0; /* Select location/length to place register */ seq.buf[1].data = data; seq.buf[1].len = 1; /* Do a polled transfer */ I2C_Status = I2C_TransferInit(i2c, &seq); while (I2C_Status == i2cTransferInProgress) { /* Enter EM1 while waiting for I2C interrupt */ EMU_EnterEM1(); /* Could do a timeout function here. */ } if (I2C_Status != i2cTransferDone) { return((int)I2C_Status); } *val = data[0]; return(0); } Page 116 Report /***************************************************************************//** * @brief * Check register content and print on LCD. * * @details * The user reads a all the set-up registers of the AD5933 and displays * the results on the screen LCD. It is intended to check that the data * stored in the AD5933 is correct. ******************************************************************************/ void AD5933_RegisterCheck(I2C_TypeDef *i2c, uint8_t addr) { uint8_t val; /* Check Control Registers */ AD5933_RegisterGet(i2c, addr, AD5933_CONTROL_15to8, &val); SegmentLCD_Write("REG0x80"); SegmentLCD_Number(val); Delay(2000); AD5933_RegisterGet(i2c, addr, AD5933_CONTROL_7to0, &val); SegmentLCD_Write("REG0x81"); SegmentLCD_Number(val); Delay(2000); /* Check Start Frequency Registers */ AD5933_RegisterGet(i2c, addr, AD5933_START_FREQUENCY_23to16, &val); SegmentLCD_Write("REG0x82"); SegmentLCD_Number(val); Delay(2000); AD5933_RegisterGet(i2c, addr, AD5933_START_FREQUENCY_15to8, &val); SegmentLCD_Write("REG0x83"); SegmentLCD_Number(val); Delay(2000); AD5933_RegisterGet(i2c, addr, AD5933_START_FREQUENCY_7to0, &val); SegmentLCD_Write("REG0x84"); SegmentLCD_Number(val); Delay(2000); /* Check Frequency Increment Registers */ AD5933_RegisterGet(i2c, addr, AD5933_FREQUENCY_INCREMENT_23to16, &val); SegmentLCD_Write("REG0x85"); SegmentLCD_Number(val); Delay(2000); AD5933_RegisterGet(i2c, addr, AD5933_FREQUENCY_INCREMENT_15to8, &val); SegmentLCD_Write("REG0x86"); SegmentLCD_Number(val); Delay(2000); AD5933_RegisterGet(i2c, addr, AD5933_FREQUENCY_INCREMENT_7to0, &val); SegmentLCD_Write("REG0x87"); SegmentLCD_Number(val); Delay(2000); /* Check Number of Increments Register */ AD5933_RegisterGet(i2c, addr, AD5933_NUM_INCREMENTS_15to8, &val); SegmentLCD_Write("REG0x88"); SegmentLCD_Number(val); Delay(2000); AD5933_RegisterGet(i2c, addr, AD5933_NUM_INCREMENTS_7to0, &val); SegmentLCD_Write("REG0x89"); SegmentLCD_Number(val); Delay(2000); Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 117 /* Check Number of Settling Cycles Register */ AD5933_RegisterGet(i2c, addr, AD5933_NUM_SETTLING_15to8, &val); SegmentLCD_Write("REG0x8A"); SegmentLCD_Number(val); Delay(2000); AD5933_RegisterGet(i2c, addr, AD5933_NUM_SETTLING_7to0, &val); SegmentLCD_Write("REG0x8B"); SegmentLCD_Number(val); Delay(2000); /* Clean LCD */ SegmentLCD_Number(false); SegmentLCD_Write(false); } /***************************************************************************//** * @brief * Program frequency sweep parameters into relevant registers. * * @details * (1) Start frequency register. * (2) Number of increments register. * (3) Frequency increment register. * Note: Refer to AD5933 Datasheet for code calculation. * * @note * User LEDs will be light up when transfer is correctly completed. * * @param[in] i2c * Pointer to I2C peripheral register block. * * @param[in] addr * I2C address for AD5933 sensor, in 8 bit format, where LSB is reserved * for R/W bit. * * @return * Returns 0 if parameters are set, <0 if unable to complete sequence. ******************************************************************************/ int AD5933_ParameterSet(I2C_TypeDef *i2c, uint8_t addr) { int status; /* Setup SysTick Timer for 1 msec interrupts */ if (SysTick_Config(CMU_ClockFreqGet(cmuClock_CORE) / 1000)) { while (1) ; } Delay(3000); //* Transmit to "Start frequency" register */ //* TBA: Description of values used */ if (AD5933_RegisterSet(i2c, addr, AD5933_START_FREQUENCY_7to0, 0x45) < 0) { status = AD5933_RegisterSet(i2c, addr, AD5933_START_FREQUENCY_7to0, 0x45); } else { BSP_LedsSet(0x8000); if (AD5933_RegisterSet(i2c, addr, AD5933_START_FREQUENCY_15to8, 0xA6) < 0) { status = AD5933_RegisterSet(i2c, addr, AD5933_START_FREQUENCY_15to8, 0xA6); } Page 118 Report else { BSP_LedsSet(0xC000); if (AD5933_RegisterSet(i2c, addr, AD5933_START_FREQUENCY_23to16, 0x0E) < 0) { status = AD5933_RegisterSet(i2c, addr, AD5933_START_FREQUENCY_23to16, 0x0E); } else { BSP_LedsSet(0xE000); //* Transmit to "Frequency increment" register */ //* TBA: Description of values used */ if (AD5933_RegisterSet(i2c, addr, AD5933_FREQUENCY_INCREMENT_7to0, 0x02) < 0) { status = AD5933_RegisterSet(i2c, addr, AD5933_FREQUENCY_INCREMENT_7to0, 0x02); } else { BSP_LedsSet(0xF000); if (AD5933_RegisterSet(i2c, addr, AD5933_FREQUENCY_INCREMENT_15to8, 0x7D) < 0) { status = AD5933_RegisterSet(i2c, addr, AD5933_FREQUENCY_INCREMENT_15to8, 0x7D); } else { BSP_LedsSet(0xF800); if (AD5933_RegisterSet(i2c, addr, AD5933_FREQUENCY_INCREMENT_23to16, 0x00) < 0) { status = AD5933_RegisterSet(i2c, addr, AD5933_FREQUENCY_INCREMENT_23to16, 0x00); } else { BSP_LedsSet(0xFC00); //* Transmit to "Number of increments" register */ //* TBA: Description of values used */ if (AD5933_RegisterSet(i2c, addr, AD5933_NUM_INCREMENTS_7to0, 0x0A) < 0) { status = AD5933_RegisterSet(i2c, addr, AD5933_NUM_INCREMENTS_7to0, 0x0A); } else { BSP_LedsSet(0xFE00); if (AD5933_RegisterSet(i2c, addr, AD5933_NUM_INCREMENTS_15to8, 0x00) < 0) { status = AD5933_RegisterSet(i2c, addr, AD5933_NUM_INCREMENTS_15to8, 0x00); } Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 119 else { BSP_LedsSet(0xFF00); //* Transmit to "Settling time cycles" register */ //* TBA: Description of values used */ if (AD5933_RegisterSet(i2c, addr, AD5933_NUM_SETTLING_7to0, 0x0F) < 0) { status = AD5933_RegisterSet(i2c, addr, AD5933_NUM_SETTLING_7to0, 0x0F); } else { BSP_LedsSet(0xFF80); if (AD5933_RegisterSet(i2c, addr, AD5933_NUM_SETTLING_15to8, 0x00) < 0) { status = AD5933_RegisterSet(i2c, addr, AD5933_NUM_SETTLING_15to8, 0x00); } else { BSP_LedsSet(0xFFC0); status = AD5933_RegisterSet(i2c, addr, AD5933_NUM_SETTLING_15to8, 0x00); } } } } } } } } } } return(status); } Page 120 Report /***************************************************************************//** * @brief * Program frequency sweep parameters into relevant registers and * place the AD5933 into standby mode, with messages for debugging. * * @details * (1) Perform AD5933_ParameterSet. * (2) Check transmission and print message accordingly. * (3) Perform AD5933_StandbyMode. * (4) Check transmission and print message accordingly. * * @note * "ERROR" message appears when there is some sort of problem. * "SET OK" and "READY" appear when the transmission is properly done. * * @param[in] i2c * Pointer to I2C peripheral register block. * * @param[in] addr * I2C address for AD5933 sensor, in 8 bit format, where LSB is reserved * for R/W bit. * ******************************************************************************/ void AD5933_Programming(I2C_TypeDef *i2c, uint8_t addr) { if (AD5933_ParameterSet(i2c, addr) < 0) { SegmentLCD_Write("ERROR"); SegmentLCD_Number(AD5933_ParameterSet(i2c, addr)); /* Enter EM2, no wakeup scheduled */ EMU_EnterEM2(true); } else { SegmentLCD_Write("SET OK"); if (AD5933_StandbyMode(i2c, addr) < 0) { SegmentLCD_Write("ERROR"); SegmentLCD_Number(AD5933_StandbyMode(i2c, addr)); /* Enter EM2, no wakeup scheduled */ EMU_EnterEM2(true); } else { SegmentLCD_Write("READY"); } } } Investigation of Lab-on-Spoon Low-Power Realization for Smart Kitchen and AAL Scenarios Page 121 /***************************************************************************//** * @brief * Place the AD5933 into standby mode. * * @details * (1) Place the AD5933 in standby mode. * (2) Choose the internal system clock. * (3) Choose range 1 (2vp-p, 1.6V) PGA = x1. * Note: Refer to AD5933 Datasheet for Control Register Map. * * @note * User LEDs will be light up when transfer is correctly completed. * * @param[in] i2c * Pointer to I2C peripheral register block. * * @param[in] addr * I2C address for AD5933 sensor, in 8 bit format, where LSB is reserved * for R/W bit. * * @return * Returns 0 if parameters are set, <0 if unable to complete sequence. ******************************************************************************/ int AD5933_StandbyMode(I2C_TypeDef *i2c, uint8_t addr) { int status; /* Setup SysTick Timer for 1 msec interrupts */ if (SysTick_Config(CMU_ClockFreqGet(cmuClock_CORE) / 1000)) { while (1) ; } Delay(1000); //* Transmit to "Control" register */ //* Place the AD5933 in standby mode */ if (AD5933_RegisterSet(i2c, addr, AD5933_CONTROL_15to8, 0xB0) < 0) { status = AD5933_RegisterSet(i2c, addr, AD5933_CONTROL_15to8, 0xB0); } else { BSP_LedsSet(0xFFE0); //* Choose the internal system clock */ if (AD5933_RegisterSet(i2c, addr, AD5933_CONTROL_7to0, 0x00) < 0) { status = AD5933_RegisterSet(i2c, addr, AD5933_CONTROL_7to0, 0x00); } else { BSP_LedsSet(0xFFF0); //* Choose range 1 (2vp-p, 1.6V) PGA = x1 */ if (AD5933_RegisterSet(i2c, addr, AD5933_CONTROL_15to8, 0x01) < 0) { status = AD5933_RegisterSet(i2c, addr, AD5933_CONTROL_15to8, 0x01); } else { BSP_LedsSet(0xFFF8); status = AD5933_RegisterSet(i2c, addr, AD5933_CONTROL_15to8, 0x01); } } } return(status); }